SAM 2.1 Desktop
Select one object in a still image with positive/negative clicks or a bounding box, returning a binary foreground mask and candidate scores.
Desktop app
Run SAM 2.1 Desktop on your computer
Download the desktop app to load this model, provide input and view results locally.
Public desktop downloads are coming soon.Device requirements
If your device falls short
MemoryAdvisory only: close other apps, reduce image/tile/token/batch settings or choose a smaller verified model/precision. CPU can need more RAM than GPU.
GPUIf the selected GPU runtime fails, existing CPU fallback may recover when enabled, but CPU has its own memory/runtime needs.
PlatformUse a compatible OS/architecture/runtime or another model. These estimates do not add hardware detection or block runs.
Benchmark evidence
Saved desktop measurements apply to the recorded hardware and runtime. Fallback cases report the device that actually ran; they do not prove GPU support on the requested device.
SAM 2.1 Small — Object Selection Desktop
77 configurations testedQuality criteria
No explicit quality gate recorded. Review per-case evidence; full certification is pending.
| Precision | Device | Result | Platform | Inference / load | |
|---|---|---|---|---|---|
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 1915 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 763 ms / 1915 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 131 ms / 1915 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1915 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 124 ms / 1915 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1915 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1915 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1915 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1915 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1987 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 42 ms / 1987 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 123 ms / 1987 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1987 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1871 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 42 ms / 1871 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 123 ms / 1871 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1871 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 2022 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 787 ms / 2022 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 121 ms / 2022 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 2022 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 117 ms / 2022 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 2022 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 2022 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 2022 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 2022 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 5 ms / 1954 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 45 ms / 1954 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 125 ms / 1954 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1954 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1930 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 44 ms / 1930 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 120 ms / 1930 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 1930 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 2033 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 852 ms / 2033 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 153 ms / 2033 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 2033 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 159 ms / 2033 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 2033 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 2033 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 2033 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 2033 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 2132 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 49 ms / 2132 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 146 ms / 2132 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 2132 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 2109 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 53 ms / 2109 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 150 ms / 2109 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 2109 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 20 ms / 1860 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 20 ms / 1860 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 583 ms / 1860 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 19 ms / 1860 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 577 ms / 1860 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 19 ms / 1860 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 20 ms / 1860 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 21 ms / 1860 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 20 ms / 1860 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 16 ms / 1964 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 1964 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 456 ms / 1964 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 1964 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 455 ms / 1964 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 1964 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 1964 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 1964 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 1964 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 1861 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 1861 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 465 ms / 1861 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 16 ms / 1861 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 17 ms / 2007 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 16 ms / 2007 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 466 ms / 2007 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 2007 ms |
Test setup & sample details
- Quality gate
- No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
- Tested
- 2026-11-01T15:18:38.358560+00:00
- Environment
- { "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2" }
- Runtime
- transformers 5.22.1
- Artifact revision
- ee5bba1d82bb8749febdf90f45e84b687142ba03
- Fixture
{ "file": "model-samples/modnet-portrait-pexels-5965592.jpg", "size": 78124, "sha1": "38212739a4a4f0be25d84863cf991d7fb6c6d818", "sha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe", "url": "https://assets.localai.tools/samples/v1/files/2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe/modnet-portrait-pexels-5965592.jpg" }- Rerun when
- artifact/runtime/processor/backend/driver; fixture checksum, prompt coordinates, expected probes/area gate; mask candidate selection/postprocessing and image-embedding cache semantics
Per-case quality & failure details
[
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1859.949333998884,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.364416998854722,
"inferenceMs": 19.708957999682752,
"masksPerSecond": 50.738349537103716,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1936.8437089979125,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25048470671669104,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1859.949333998884,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.99374999954307,
"inferenceMs": 19.833042000755086,
"masksPerSecond": 50.42090870184855,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1938.7155429994891,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37545432037633075,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1859.949333998884,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 588.0286660012644,
"inferenceMs": 582.7375420012686,
"masksPerSecond": 1.7160384013800556,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2480.664833998162,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37256746719485023,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1859.949333998884,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.84787499997765,
"inferenceMs": 19.292791999760084,
"masksPerSecond": 51.832829587984754,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1918.3311679989856,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1859.949333998884,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 582.3104999999487,
"inferenceMs": 576.7224589999387,
"masksPerSecond": 1.733936288408193,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2492.616500998338,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1859.949333998884,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.477916000250843,
"inferenceMs": 19.04933399964648,
"masksPerSecond": 52.49527358901672,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1937.3619999987568,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1859.949333998884,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.214583999011666,
"inferenceMs": 20.175499999822932,
"masksPerSecond": 49.56506654153683,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1945.8964169989486,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1859.949333998884,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.61037500022212,
"inferenceMs": 21.434250000311295,
"masksPerSecond": 46.654303275620876,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1938.1289999982982,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1859.949333998884,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.489875000857865,
"inferenceMs": 19.94583300074737,
"masksPerSecond": 50.13578525211406,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1937.1168759989814,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1963.654041999689,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.11983299962594,
"inferenceMs": 16.32762500048557,
"masksPerSecond": 61.24589460930545,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2036.871374999464,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25048470671669104,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1963.654041999689,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.243041000663652,
"inferenceMs": 15.390874999866355,
"masksPerSecond": 64.97356388176003,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2033.9230000008683,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37545432037633075,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1963.654041999689,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 459.5602500012319,
"inferenceMs": 455.7931250001275,
"masksPerSecond": 2.1939778051714147,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2457.0771669987153,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37256746719485023,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1963.654041999689,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.453041000611847,
"inferenceMs": 14.473416000328143,
"masksPerSecond": 69.09218943042389,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2016.5305419996002,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1963.654041999689,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 458.88074999857054,
"inferenceMs": 455.20562499950756,
"masksPerSecond": 2.196809408937075,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2472.44987600061,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1963.654041999689,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.980749999172986,
"inferenceMs": 15.178625000771717,
"masksPerSecond": 65.88212041269598,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2035.8172920005018,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1963.654041999689,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.047708001176943,
"inferenceMs": 14.269749999584747,
"masksPerSecond": 70.07831251627395,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2033.0992090002837,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1963.654041999689,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.313958000362618,
"inferenceMs": 14.400542000657879,
"masksPerSecond": 69.44183072792092,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2032.2326259993133,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1963.654041999689,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.01075000023411,
"inferenceMs": 15.05133299906447,
"masksPerSecond": 66.43929810483603,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2032.9430419988057,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1860.6577919999836,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.37033300055191,
"inferenceMs": 15.436624998983461,
"masksPerSecond": 64.78099973704435,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1930.1120830004947,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25048470671669104,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1860.6577919999836,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.979209000463015,
"inferenceMs": 15.123790999496123,
"masksPerSecond": 66.12098778892917,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1931.1882509991847,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37545432037633075,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1860.6577919999836,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 469.7404579983413,
"inferenceMs": 465.4766669991659,
"masksPerSecond": 2.1483353965877563,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2364.64504200012,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37256746719485023,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1860.6577919999836,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.43424999990384,
"inferenceMs": 15.657291000024998,
"masksPerSecond": 63.86800883999687,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 1914.3859169998905,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25620967926427524,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2006.5342079997208,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.24466600071173,
"inferenceMs": 17.15962499838497,
"masksPerSecond": 58.27633180177993,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2080.9860409990506,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25048470671669104,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2006.5342079997208,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.406124998771702,
"inferenceMs": 16.390417000366142,
"masksPerSecond": 61.011260419894214,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2078.4470830003556,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37545432037633075,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2006.5342079997208,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 469.9122920010268,
"inferenceMs": 466.3403329996072,
"masksPerSecond": 2.1443566623709605,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2510.8796659988,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37256746719485023,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2006.5342079997208,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.998166000325,
"inferenceMs": 14.447957999436767,
"masksPerSecond": 69.2139332104221,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2059.4204169992736,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2549299963396779,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1914.8507500012784,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.816666000522673,
"inferenceMs": 9.542290999888792,
"masksPerSecond": 104.7966363645433,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2129854464,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 1984.282584000539,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24509431048682284,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1914.8507500012784,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 774.3780000000697,
"inferenceMs": 763.0739999985963,
"masksPerSecond": 1.3104888909880819,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 2739.172542002052,
"grade": "B"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3706189650903689,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1914.8507500012784,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 137.47212500129535,
"inferenceMs": 130.83470799938368,
"masksPerSecond": 7.643231794461686,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 2086.2757910017535,
"grade": "B"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3717058677890567,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1914.8507500012784,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.846749999880558,
"inferenceMs": 6.0588329997699475,
"masksPerSecond": 165.04828570749018,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 1961.6892500016547,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2549299963396779,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1914.8507500012784,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 131.37941600143677,
"inferenceMs": 123.50237499958894,
"masksPerSecond": 8.097010280193627,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 2096.944375000021,
"grade": "B"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2549299963396779,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1914.8507500012784,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.853999998696963,
"inferenceMs": 5.572500000198488,
"masksPerSecond": 179.45266935206473,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 1980.0662090019614,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2549299963396779,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1914.8507500012784,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.578874999438995,
"inferenceMs": 6.046708000212675,
"masksPerSecond": 165.37924436980055,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 1978.6200830003509,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2549299963396779,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1914.8507500012784,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.850957999238744,
"inferenceMs": 6.030083000950981,
"masksPerSecond": 165.83519660381023,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 1980.7870000004186,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2549299963396779,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1914.8507500012784,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.6239169985638,
"inferenceMs": 6.117958999311668,
"masksPerSecond": 163.45320393819412,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 1977.6120000005903,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2549299963396779,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1987.0612920003623,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.27879199925519,
"inferenceMs": 5.987124999592197,
"masksPerSecond": 167.02507465070687,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2129870848,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 2049.944833999689,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24509431048682284,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1987.0612920003623,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 49.45591700015939,
"inferenceMs": 42.30345799987845,
"masksPerSecond": 23.638729486437573,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 2085.519751000902,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3706189650903689,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1987.0612920003623,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 129.52016700000968,
"inferenceMs": 122.73875000028056,
"masksPerSecond": 8.147386216640744,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 2149.5046250001906,
"grade": "B"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3717058677890567,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1987.0612920003623,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.153583999155671,
"inferenceMs": 5.7878339994204,
"masksPerSecond": 172.77620610752504,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 2032.6524580013938,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2549299963396779,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1871.421833999193,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.130999999499181,
"inferenceMs": 5.7043750002776505,
"masksPerSecond": 175.3040429409579,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2129870848,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 1935.46916699961,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24509431048682284,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1871.421833999193,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 49.79349999848637,
"inferenceMs": 42.04712499995367,
"masksPerSecond": 23.782838898048368,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 1969.5849589988939,
"grade": "A"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3706189650903689,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1871.421833999193,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 130.84091699965938,
"inferenceMs": 123.38199999976496,
"masksPerSecond": 8.104909954465846,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 2034.6308759981184,
"grade": "B"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3717058677890567,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1871.421833999193,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 355
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.28566600002523,
"inferenceMs": 6.137667000075453,
"masksPerSecond": 162.92835697793748,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "bf16",
"totalMs": 1916.4915429992107,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25490140007320644,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2021.5851659995678,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.95679099955305,
"inferenceMs": 8.093833999737399,
"masksPerSecond": 123.55084129875219,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2129854464,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2087.3405410002306,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24517009059297218,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2021.5851659995678,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 795.2995830000873,
"inferenceMs": 786.6700419999688,
"masksPerSecond": 1.271180986449767,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2866.177083000366,
"grade": "B"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3706412478336222,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2021.5851659995678,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 127.58337499872141,
"inferenceMs": 120.73345899989363,
"masksPerSecond": 8.282708109943915,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2184.1482910003833,
"grade": "B"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37187917801436,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2021.5851659995678,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.674167001227033,
"inferenceMs": 5.702667000150541,
"masksPerSecond": 175.35654808067903,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2069.8467079982947,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25490140007320644,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2021.5851659995678,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 124.39741600064735,
"inferenceMs": 117.21125000076427,
"masksPerSecond": 8.531604261480698,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2195.716790998631,
"grade": "B"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25490140007320644,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2021.5851659995678,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.32966700081306,
"inferenceMs": 6.5994579999824055,
"masksPerSecond": 151.5275951453386,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2083.467415999621,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25490140007320644,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2021.5851659995678,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.852207999458187,
"inferenceMs": 5.599541000265162,
"masksPerSecond": 178.58606624233053,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2086.4688749989,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25490140007320644,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2021.5851659995678,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.122083000780549,
"inferenceMs": 5.931458001214196,
"masksPerSecond": 168.59261243952085,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2089.381081999818,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25490140007320644,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2021.5851659995678,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.224000000263914,
"inferenceMs": 6.587416999536799,
"masksPerSecond": 151.80456923712526,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135867392,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2085.5507500000385,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25490140007320644,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1954.3428329998278,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 11.527208998813876,
"inferenceMs": 5.438750000394066,
"masksPerSecond": 183.86577796875105,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2129870848,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2014.6317489998182,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24517009059297218,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1954.3428329998278,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 52.981582999564125,
"inferenceMs": 44.99829099950148,
"masksPerSecond": 22.223066205138295,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2056.831083000361,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3706412478336222,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1954.3428329998278,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 130.22991600155365,
"inferenceMs": 124.59387499984587,
"masksPerSecond": 8.026076723283847,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2118.9194159996987,
"grade": "B"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37187917801436,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1954.3428329998278,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.132916999689769,
"inferenceMs": 6.2450000004901085,
"masksPerSecond": 160.1281024694187,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 1999.260458000208,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25490140007320644,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1929.9220839984628,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.530040998375625,
"inferenceMs": 5.860249999386724,
"masksPerSecond": 170.64118426767632,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2129870848,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 1995.3057089987851,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24517009059297218,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1929.9220839984628,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 52.37670799942862,
"inferenceMs": 43.84837500037975,
"masksPerSecond": 22.805862246693053,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2032.1882919979544,
"grade": "A"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3706412478336222,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1929.9220839984628,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 127.9748329998256,
"inferenceMs": 120.43908400119108,
"masksPerSecond": 8.302952553094064,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 2091.7145009989326,
"grade": "B"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.37187917801436,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 1929.9220839984628,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 355
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.17312500011758,
"inferenceMs": 6.392999999661697,
"masksPerSecond": 156.42108557061124,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2135883776,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp16",
"totalMs": 1975.463166998452,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.254892821193265,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2032.940208000582,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.564042000638437,
"inferenceMs": 13.103332999889972,
"masksPerSecond": 76.3164608583478,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2241003520,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2105.223123999167,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24515579245973645,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2032.940208000582,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 861.8792500001291,
"inferenceMs": 852.4709999983315,
"masksPerSecond": 1.1730604325565999,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2247016448,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2946.612040999753,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3706412478336222,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2032.940208000582,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 158.50379199946474,
"inferenceMs": 152.79279199967277,
"masksPerSecond": 6.544811354727661,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320758272,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2226.0454579991347,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3718692745729141,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2032.940208000582,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.82800000142015,
"inferenceMs": 8.112542000162648,
"masksPerSecond": 123.26592577023959,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320758272,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2081.749375000072,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.254892821193265,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2032.940208000582,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 168.3983340008126,
"inferenceMs": 158.95266699953936,
"masksPerSecond": 6.291181009267985,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320758272,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2282.3851660014043,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.254892821193265,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2032.940208000582,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.824000000066007,
"inferenceMs": 7.593667000037385,
"masksPerSecond": 131.68868216042088,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320758272,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2100.436916000035,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.254892821193265,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2032.940208000582,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 11.67008300035377,
"inferenceMs": 6.254083000385435,
"masksPerSecond": 159.89554342952766,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320758272,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2094.6426240006986,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.254892821193265,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2032.940208000582,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.907582999716396,
"inferenceMs": 8.02220799960196,
"masksPerSecond": 124.6539606115445,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320758272,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2101.0073329998704,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.254892821193265,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2032.940208000582,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.338582999902428,
"inferenceMs": 8.335958000316168,
"masksPerSecond": 119.96221669567814,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320758272,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2101.045667001017,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.254892821193265,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2132.0418329996755,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.735916000063298,
"inferenceMs": 7.641166001121746,
"masksPerSecond": 130.87007923309048,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2241019904,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2198.316457999681,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24515579245973645,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2132.0418329996755,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 56.50975000025937,
"inferenceMs": 49.21325000032084,
"masksPerSecond": 20.31973096662953,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2247032832,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2241.4244579995284,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3706412478336222,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2132.0418329996755,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 151.81658400069864,
"inferenceMs": 145.75083300042024,
"masksPerSecond": 6.8610242522395515,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320774656,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2317.91316700037,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3718692745729141,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2132.0418329996755,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.444291000472731,
"inferenceMs": 7.327291001274716,
"masksPerSecond": 136.47608643167464,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320774656,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2181.425541999488,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.254892821193265,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2108.6301659997844,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.084750000823988,
"inferenceMs": 7.922708999103634,
"masksPerSecond": 126.21945348657118,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2241019904,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2177.5373330001457,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24515579245973645,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2108.6301659997844,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 60.886875000505825,
"inferenceMs": 52.67233300037333,
"masksPerSecond": 18.985299170114835,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2247032832,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2221.639998999308,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3706412478336222,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2108.6301659997844,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 155.70970899898384,
"inferenceMs": 149.80841699980374,
"masksPerSecond": 6.675192355856147,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320774656,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2300.1885829999082,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3718692745729141,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 2108.6301659997844,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 355
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.838374999773805,
"inferenceMs": 7.017583999186172,
"masksPerSecond": 142.49918492118795,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3320774656,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"artifactRevision": "ee5bba1d82bb8749febdf90f45e84b687142ba03",
"precision": "fp32",
"totalMs": 2157.522333000088,
"grade": "A"
}
}
]SAM2.1 Base Plus — Object Selection Desktop
36 configurations testedQuality criteria
No explicit quality gate recorded. Review per-case evidence; full certification is pending.
| Precision | Device | Result | Platform | Inference / load | |
|---|---|---|---|---|---|
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 1494 ms / 296 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 44 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 245 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 245 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 391 ms / 301 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 6 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 370 ms / 288 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1734 ms / 108 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1639 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1693 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1708 ms / 106 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1703 ms / 108 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 861 ms / 105 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 813 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 805 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 888 ms / 108 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 879 ms / 112 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
Test setup & sample details
- Quality gate
- No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
- Tested
- 2026-11-01T21:51:15.221976+00:00
- Environment
- { "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2", "python": "app-owned CPython3.13.18 private probe venv", "runtime": "Transformers5.22.1/Torch2.10.0" }
- Runtime
- transformers 5.22.1
- Artifact revision
- b7320756a13354e7530a63935656d35b2f91a290
- Fixture
{ "file": "model-samples/modnet-portrait-pexels-5965592.jpg", "size": 78124, "sha1": "38212739a4a4f0be25d84863cf991d7fb6c6d818", "sha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe", "url": "https://assets.localai.tools/samples/v1/files/2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe/modnet-portrait-pexels-5965592.jpg" }- Rerun when
- artifact/runtime/processor/backend/driver; fixture checksum, prompt coordinates, expected probes/area gate; mask candidate selection/postprocessing and image-embedding cache semantics; Concrete variant source/artifact/dimension/architecture change
Per-case quality & failure details
[
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 108.33762499532895,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1748.5738749965094,
"inferenceMs": 1734.1319589977502,
"masksPerSecond": 0.5766573845844781,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 1856.9114999918384,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "C"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2520560715592972,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.44412499933969,
"inferenceMs": 29.56666699901689,
"masksPerSecond": 33.82187109670666,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 40.44412499933969,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3744590245110176,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1648.9434159957455,
"inferenceMs": 1639.1242500030785,
"masksPerSecond": 0.6100818775624373,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 1648.9434159957455,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "C"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3711983164149542,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.917207999678794,
"inferenceMs": 29.987582995090634,
"masksPerSecond": 33.347135718264234,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 39.917207999678794,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1704.0027080001892,
"inferenceMs": 1693.3440420034458,
"masksPerSecond": 0.5905474464698091,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 1704.0027080001892,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "C"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.701458998024464,
"inferenceMs": 28.808750001189765,
"masksPerSecond": 34.711676138628064,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 39.701458998024464,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 106.30925000441493,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1722.2450000044773,
"inferenceMs": 1707.764083002985,
"masksPerSecond": 0.5855609741139239,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 1828.5542500088923,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "C"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.61200000048848,
"inferenceMs": 29.025292002188507,
"masksPerSecond": 34.45271110191071,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 39.61200000048848,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.61916700063739,
"inferenceMs": 28.639416996156797,
"masksPerSecond": 34.916911895734216,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 39.61916700063739,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 108.18145800294587,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1717.606167003396,
"inferenceMs": 1702.9466669991962,
"masksPerSecond": 0.5872174504220525,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 1825.7876250063418,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "C"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.06008299620589,
"inferenceMs": 29.10041699942667,
"masksPerSecond": 34.36376873979853,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 40.06008299620589,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.82720800215611,
"inferenceMs": 30.014458003279287,
"masksPerSecond": 33.31727662351068,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 40.82720800215611,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 104.75887500069803,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 867.9661249989294,
"inferenceMs": 861.1845829946105,
"masksPerSecond": 1.1611912472035721,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 972.7249999996275,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2520560715592972,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.890166003780905,
"inferenceMs": 14.13558299827855,
"masksPerSecond": 70.74345643344044,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.890166003780905,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3744590245110176,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 816.6589170068619,
"inferenceMs": 813.176374998875,
"masksPerSecond": 1.2297455149276606,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 816.6589170068619,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3711983164149542,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.35120899725007,
"inferenceMs": 15.779250003106426,
"masksPerSecond": 63.374368224290265,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 19.35120899725007,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 808.5278340004152,
"inferenceMs": 804.7191250007018,
"masksPerSecond": 1.2426696084787694,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 808.5278340004152,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.756333996658213,
"inferenceMs": 13.997957998071797,
"masksPerSecond": 71.43899132557398,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.756333996658213,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 108.49504199723015,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 894.7889170012786,
"inferenceMs": 887.7297500002896,
"masksPerSecond": 1.1264689507135182,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 1003.2839589985088,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.806333999847993,
"inferenceMs": 17.385374994773883,
"masksPerSecond": 57.51961060952693,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 21.806333999847993,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.799916004762053,
"inferenceMs": 16.975875005300622,
"masksPerSecond": 58.9071255347813,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 20.799916004762053,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 112.17620800016448,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 885.9904580021976,
"inferenceMs": 878.8969579982222,
"masksPerSecond": 1.1377898067568721,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 998.166666002362,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.241416993783787,
"inferenceMs": 13.611834001494572,
"masksPerSecond": 73.46548598008178,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.241416993783787,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.378417003783397,
"inferenceMs": 13.520333006454166,
"masksPerSecond": 73.96267529228996,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.378417003783397,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 295.9757080025156,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1505.5507500001113,
"inferenceMs": 1494.4776670017745,
"masksPerSecond": 0.6691301061769648,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3705323520,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 1801.526458002627,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "C"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2520560715592972,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 52.17616700247163,
"inferenceMs": 44.10458299389575,
"masksPerSecond": 22.673380681059925,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3715530752,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 52.17616700247163,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3744590245110176,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 252.6626250037225,
"inferenceMs": 245.27504100115038,
"masksPerSecond": 4.077055683767309,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4252401664,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 252.6626250037225,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3711983164149542,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.527499995892867,
"inferenceMs": 6.891874996654224,
"masksPerSecond": 145.09839491944743,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4252401664,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 14.527499995892867,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 251.79183399450267,
"inferenceMs": 244.84108300384833,
"masksPerSecond": 4.084281884933022,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4252401664,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 251.79183399450267,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.215041999705136,
"inferenceMs": 7.08670899621211,
"masksPerSecond": 141.10922298834427,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4252401664,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 14.215041999705136,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 300.6952910000109,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 401.94666699972004,
"inferenceMs": 391.28612499916926,
"masksPerSecond": 2.5556745719059757,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3705339904,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 702.641957999731,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.555999998061452,
"inferenceMs": 5.680083995684981,
"masksPerSecond": 176.05373455034734,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3709534208,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 13.555999998061452,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.394207999226637,
"inferenceMs": 6.405790998542216,
"masksPerSecond": 156.1087460124086,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3709534208,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 13.394207999226637,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 288.03604200220434,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 378.3794580012909,
"inferenceMs": 369.6095420018537,
"masksPerSecond": 2.705557855957584,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3705339904,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 666.4155000034953,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.592375002393965,
"inferenceMs": 7.21266699838452,
"masksPerSecond": 138.64497005393125,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3709534208,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 14.592375002393965,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25608385569180087,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 451
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.756084001623094,
"inferenceMs": 7.5238330027787015,
"masksPerSecond": 132.91097764007787,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3709534208,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 15.756084001623094,
"artifactRevision": "b7320756a13354e7530a63935656d35b2f91a290",
"fallback": false,
"grade": "A"
}
}
]SAM 2.1 Large — Image Object Selection Desktop
364 configurations testedQuality criteria
No explicit quality gate recorded. Review per-case evidence; full certification is pending.
| Precision | Device | Result | Platform | Inference / load | |
|---|---|---|---|---|---|
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 1347 ms / 455 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 61 ms / 455 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 846 ms / 455 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 455 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 1134 ms / 459 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 56 ms / 459 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 811 ms / 459 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 459 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 1006 ms / 441 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 49 ms / 441 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 734 ms / 441 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 441 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 4992 ms / 380 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 22 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 22 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 20 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 791 ms / 380 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 24 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 21 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 22 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 607 ms / 380 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 20 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 380 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 1107 ms / 352 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 22 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 65 ms / 352 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 22 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 24 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 21 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 944 ms / 352 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 20 ms / 352 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 23 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 28 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 20 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 1105 ms / 364 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 71 ms / 364 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 21 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 26 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 21 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 694 ms / 364 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 23 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 20 ms / 364 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 21 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 22 ms / 0 ms |
FP16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 5322 ms / 368 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 20 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 24 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 788 ms / 368 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 26 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 21 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 20 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 927 ms / 368 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 368 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 1145 ms / 348 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 24 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 61 ms / 348 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 24 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 20 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 1009 ms / 348 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 348 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 13 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 1127 ms / 374 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 16 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 55 ms / 374 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 984 ms / 374 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 11 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 12 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 14 ms / 374 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 18 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 17 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 19 ms / 0 ms |
BF16 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 2630 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 10 ms / 2630 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 8 ms / 2630 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 9 ms / 2630 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 3185 ms / 150 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 150 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 32 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 32 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 3030 ms / 150 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 32 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 32 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 150 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 3168 ms / 149 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 149 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 33 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 3084 ms / 149 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 149 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 33 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 3176 ms / 162 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 33 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 32 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 162 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 34 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 3050 ms / 162 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 28 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 162 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 31 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 29 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 30 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1821 ms / 149 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 149 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1769 ms / 149 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 149 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1803 ms / 152 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 152 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1711 ms / 152 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 152 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1791 ms / 159 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 16 ms / 159 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 1693 ms / 159 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 159 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
Test setup & sample details
- Quality gate
- No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
- Tested
- 2026-11-05T23:51:41.106806+00:00
- Environment
- { "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2", "ramBytes": 103079215104 }
- Runtime
- transformers 5.22.1
- Artifact revision
- 665f8e2ad61cf5f53d65644ff27c8ee525124610
- Fixture
{ "file": "model-samples/modnet-portrait-pexels-5965592.jpg", "size": 78124, "sha1": "38212739a4a4f0be25d84863cf991d7fb6c6d818", "url": "https://assets.localai.tools/samples/v1/files/2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe/modnet-portrait-pexels-5965592.jpg" }- Rerun when
- artifact/runtime/processor/backend/driver; fixture checksum, prompt coordinates, expected probes/area gate; mask candidate selection/postprocessing and image-embedding cache semantics
Per-case quality & failure details
[
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 454.51141701778397,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1364.5177500438876,
"inferenceMs": 1347.138500015717,
"masksPerSecond": 0.7423141718452357,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4716822528,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1819.0291670616716,
"processAndInitWallMs": 2069.466040993575,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.519168593275183
},
"observedPeakRssBytes": 488718336
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.500582976732403,
"inferenceMs": 10.95516700297594,
"masksPerSecond": 91.28112786672743,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.500582976732403,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 63.84129058096624
},
"observedPeakRssBytes": 528449536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.04833302553743,
"inferenceMs": 10.51350001944229,
"masksPerSecond": 95.11580331485528,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.04833302553743,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 66.52323191198326
},
"observedPeakRssBytes": 528449536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.756542005576193,
"inferenceMs": 12.341499968897551,
"masksPerSecond": 81.02742798850637,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.756542005576193,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 56.66993491573745
},
"observedPeakRssBytes": 528449536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.866125017870218,
"inferenceMs": 14.777332951780409,
"masksPerSecond": 67.6712099039169,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.866125017870218,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.328702837120254
},
"observedPeakRssBytes": 528449536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.443083030637354,
"inferenceMs": 12.91874999878928,
"masksPerSecond": 77.40686986695447,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.443083030637354,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 54.13774552998902
},
"observedPeakRssBytes": 528449536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 454.51141701778397,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 69.39579104073346,
"inferenceMs": 60.606167011428624,
"masksPerSecond": 16.499971031189418,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 523.9072080585174,
"processAndInitWallMs": 2069.466040993575,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 11.53994773944563
},
"observedPeakRssBytes": 528449536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.751374995801598,
"inferenceMs": 16.482667007949203,
"masksPerSecond": 60.66979327542827,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.751374995801598,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.43196805848833
},
"observedPeakRssBytes": 560726016
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.93962498474866,
"inferenceMs": 15.457624976988882,
"masksPerSecond": 64.69299141935828,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 20.93962498474866,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 45.24576065476783
},
"observedPeakRssBytes": 560726016
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.271041023079306,
"inferenceMs": 15.013333992101252,
"masksPerSecond": 66.60745711286484,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 20.271041023079306,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 46.58472264508077
},
"observedPeakRssBytes": 560726016
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.3984590373002,
"inferenceMs": 13.694000022951514,
"masksPerSecond": 73.02468222024048,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.3984590373002,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 51.07287854737842
},
"observedPeakRssBytes": 560726016
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.70925003197044,
"inferenceMs": 13.979208015371114,
"masksPerSecond": 71.53481076327287,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.70925003197044,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 50.03087436934694
},
"observedPeakRssBytes": 560726016
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 454.51141701778397,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 852.0224589738064,
"inferenceMs": 845.5038749962114,
"masksPerSecond": 1.1827266906428795,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1306.5338759915903,
"processAndInitWallMs": 2069.466040993575,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.47770331035065905
},
"observedPeakRssBytes": 577421312
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.659916003234684,
"inferenceMs": 8.51120799779892,
"masksPerSecond": 117.49213510686258,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 13.659916003234684,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.45507336966179
},
"observedPeakRssBytes": 577421312
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.047417022287846,
"inferenceMs": 9.397791989613324,
"masksPerSecond": 106.40797339473198,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.047417022287846,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.97818045413225
},
"observedPeakRssBytes": 577421312
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.97212499473244,
"inferenceMs": 9.958290960639715,
"masksPerSecond": 100.41883732384544,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.97212499473244,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 40.55916839510117
},
"observedPeakRssBytes": 581861376
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.59312500851229,
"inferenceMs": 11.220584041438997,
"masksPerSecond": 89.12192059761568,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.59312500851229,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 35.996343729376974
},
"observedPeakRssBytes": 581861376
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.90370902745053,
"inferenceMs": 12.210417015012354,
"masksPerSecond": 81.89728481595093,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.90370902745053,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 33.07831333716258
},
"observedPeakRssBytes": 581861376
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 454.51141701778397,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.281208991538733,
"inferenceMs": 11.796833015978336,
"masksPerSecond": 84.76851360407834,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 472.7926260093227,
"processAndInitWallMs": 2069.466040993575,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 34.238002644687235
},
"observedPeakRssBytes": 581861376
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.76379200257361,
"inferenceMs": 11.544874985702336,
"masksPerSecond": 86.61852131256879,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.76379200257361,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 34.98522075814653
},
"observedPeakRssBytes": 581861376
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.110374978277832,
"inferenceMs": 10.856167005840689,
"masksPerSecond": 92.11354241897655,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.110374978277832,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 37.20465978302463
},
"observedPeakRssBytes": 581861376
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.766250006388873,
"inferenceMs": 12.142333027441055,
"masksPerSecond": 82.35649588427948,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.766250006388873,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 33.26378868766048
},
"observedPeakRssBytes": 581861376
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.88341601099819,
"inferenceMs": 14.369499986059964,
"masksPerSecond": 69.59184390341437,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.88341601099819,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.108145752589063
},
"observedPeakRssBytes": 581861376
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.117542007938027,
"inferenceMs": 13.01695901202038,
"masksPerSecond": 76.82285847843265,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 20.117542007938027,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 31.028752539438944
},
"observedPeakRssBytes": 581877760
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 458.9529170189053,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1142.590458970517,
"inferenceMs": 1134.4234580174088,
"masksPerSecond": 0.8815050437582312,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4716822528,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1601.5433759894222,
"processAndInitWallMs": 2018.7438330031,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.6165175755641569
},
"observedPeakRssBytes": 492404736
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.730458031408489,
"inferenceMs": 9.353707951959223,
"masksPerSecond": 106.90947431072408,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.730458031408489,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 74.77163105712593
},
"observedPeakRssBytes": 492404736
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.233833983074874,
"inferenceMs": 9.743917034938931,
"masksPerSecond": 102.62813162450817,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.233833983074874,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 71.77729423312802
},
"observedPeakRssBytes": 492404736
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.19862502068281,
"inferenceMs": 11.52420899597928,
"masksPerSecond": 86.7738514937462,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.19862502068281,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 60.68893754391414
},
"observedPeakRssBytes": 492404736
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.462708950508386,
"inferenceMs": 9.861832950264215,
"masksPerSecond": 101.40102808912498,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.462708950508386,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 70.9190678373093
},
"observedPeakRssBytes": 540688384
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.524249978829175,
"inferenceMs": 11.260333994869143,
"masksPerSecond": 88.80731250561996,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.524249978829175,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 62.111123907930555
},
"observedPeakRssBytes": 540688384
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 458.9529170189053,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 63.778874988202006,
"inferenceMs": 56.33437499636784,
"masksPerSecond": 17.75115105234548,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 522.7317920071073,
"processAndInitWallMs": 2018.7438330031,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 12.41501303680201
},
"observedPeakRssBytes": 540688384
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.591874977573752,
"inferenceMs": 10.963917011395097,
"masksPerSecond": 91.20827884420073,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.591874977573752,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 63.79034055740325
},
"observedPeakRssBytes": 540688384
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.247750031761825,
"inferenceMs": 12.982583022676408,
"masksPerSecond": 77.0262742208789,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.247750031761825,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 53.87155997988894
},
"observedPeakRssBytes": 540688384
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.410874974913895,
"inferenceMs": 14.510541048366576,
"masksPerSecond": 68.91541787909887,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 20.410874974913895,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.19889194129872
},
"observedPeakRssBytes": 567721984
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.855124981608242,
"inferenceMs": 15.0167919928208,
"masksPerSecond": 66.59211904100944,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 20.855124981608242,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 46.57399532032967
},
"observedPeakRssBytes": 567721984
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.63595804106444,
"inferenceMs": 16.86633296776563,
"masksPerSecond": 59.28971056786123,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.63595804106444,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 41.4667492534776
},
"observedPeakRssBytes": 567721984
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 458.9529170189053,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 818.1505419779569,
"inferenceMs": 810.7251250185072,
"masksPerSecond": 1.2334636847194935,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1277.1034589968622,
"processAndInitWallMs": 2018.7438330031,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.49819598225820344
},
"observedPeakRssBytes": 578306048
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.430332987103611,
"inferenceMs": 8.460666984319687,
"masksPerSecond": 118.1939913074606,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.430332987103611,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.73855308908334
},
"observedPeakRssBytes": 578306048
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.61954100523144,
"inferenceMs": 9.009957953821868,
"masksPerSecond": 110.98830928237766,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.61954100523144,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 44.828178119152334
},
"observedPeakRssBytes": 578306048
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.751499984413385,
"inferenceMs": 9.008207998704165,
"masksPerSecond": 111.00987012554,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.751499984413385,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 44.8368865437056
},
"observedPeakRssBytes": 578306048
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.509792032185942,
"inferenceMs": 8.570708974730223,
"masksPerSecond": 116.67646199963015,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.509792032185942,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.12562300165062
},
"observedPeakRssBytes": 578306048
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.26279200334102,
"inferenceMs": 9.247416979633272,
"masksPerSecond": 108.13830523728123,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.26279200334102,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 43.67706148533789
},
"observedPeakRssBytes": 578306048
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 458.9529170189053,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.22704097442329,
"inferenceMs": 9.988292003981769,
"masksPerSecond": 100.11721719803108,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 474.17995799332857,
"processAndInitWallMs": 2018.7438330031,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 40.437344026284755
},
"observedPeakRssBytes": 578306048
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.015458037145436,
"inferenceMs": 11.05758297489956,
"masksPerSecond": 90.43567679030538,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.015458037145436,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 36.526969855604335
},
"observedPeakRssBytes": 578306048
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.038416990544647,
"inferenceMs": 10.981624945998192,
"masksPerSecond": 91.0612049598734,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.038416990544647,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 36.77962068329286
},
"observedPeakRssBytes": 581697536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.518541960977018,
"inferenceMs": 11.207957984879613,
"masksPerSecond": 89.22231876217559,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.518541960977018,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 36.03689454804272
},
"observedPeakRssBytes": 581697536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.49270800407976,
"inferenceMs": 10.900542023591697,
"masksPerSecond": 91.73855738877312,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.49270800407976,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 37.05320332932546
},
"observedPeakRssBytes": 581697536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.83245796104893,
"inferenceMs": 12.159250036347657,
"masksPerSecond": 82.24191434592586,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.83245796104893,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 33.217509204319455
},
"observedPeakRssBytes": 581697536
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 441.45958399167284,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1013.9722910244018,
"inferenceMs": 1005.5866669863462,
"masksPerSecond": 0.9944443704657611,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4716822528,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1455.4318750160746,
"processAndInitWallMs": 1996.0824169684201,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.6955064371487896
},
"observedPeakRssBytes": 492961792
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.825082966126502,
"inferenceMs": 8.822959032841027,
"masksPerSecond": 113.34066000734859,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.825082966126502,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 79.26955088385955
},
"observedPeakRssBytes": 525254656
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.850874992087483,
"inferenceMs": 10.15662495046854,
"masksPerSecond": 98.45790357296481,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.850874992087483,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 68.860670095703
},
"observedPeakRssBytes": 525254656
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.79412499954924,
"inferenceMs": 9.06837498769164,
"masksPerSecond": 110.2733401913004,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 13.79412499954924,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 77.12429194307398
},
"observedPeakRssBytes": 525254656
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.927791002672166,
"inferenceMs": 10.652167024090886,
"masksPerSecond": 93.87761173274932,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.927791002672166,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 65.65725062499102
},
"observedPeakRssBytes": 525254656
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.466958004049957,
"inferenceMs": 9.143333008978516,
"masksPerSecond": 109.36930756191707,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4721016832,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.466958004049957,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 76.49201875434431
},
"observedPeakRssBytes": 525254656
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 441.45958399167284,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 55.257792002521455,
"inferenceMs": 48.677083977963775,
"masksPerSecond": 20.543547769884945,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 496.7173759941943,
"processAndInitWallMs": 1996.0824169684201,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 14.367992961875373
},
"observedPeakRssBytes": 563544064
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.256417020689696,
"inferenceMs": 12.150958005804569,
"masksPerSecond": 82.29803769565292,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.256417020689696,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 57.55858918003809
},
"observedPeakRssBytes": 563544064
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.44541700463742,
"inferenceMs": 10.718834004364908,
"masksPerSecond": 93.29372948520161,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.44541700463742,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 65.24888805211413
},
"observedPeakRssBytes": 563544064
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.096000003628433,
"inferenceMs": 13.210667006205767,
"masksPerSecond": 75.69640499834306,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.096000003628433,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 52.941460084601154
},
"observedPeakRssBytes": 563544064
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.012791988439858,
"inferenceMs": 12.664292007684708,
"masksPerSecond": 78.96217170238958,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.012791988439858,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 55.225511191277654
},
"observedPeakRssBytes": 563544064
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.749209015164524,
"inferenceMs": 11.515958001837134,
"masksPerSecond": 86.8360235284351,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.749209015164524,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 60.732420167599294
},
"observedPeakRssBytes": 563544064
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 441.45958399167284,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 739.969625021331,
"inferenceMs": 733.5619169753045,
"masksPerSecond": 1.3632114438591627,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1181.4292090130039,
"processAndInitWallMs": 1996.0824169684201,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.5506011021747158
},
"observedPeakRssBytes": 574849024
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 12.948957970365882,
"inferenceMs": 7.715000014286488,
"masksPerSecond": 129.6176277573842,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 12.948957970365882,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 52.35255985120749
},
"observedPeakRssBytes": 574849024
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.331832993775606,
"inferenceMs": 9.010625013615936,
"masksPerSecond": 110.98009277812606,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.331832993775606,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 44.82485947308511
},
"observedPeakRssBytes": 574849024
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.138959009665996,
"inferenceMs": 8.244208001997322,
"masksPerSecond": 121.29727922412077,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 13.138959009665996,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.99197107862238
},
"observedPeakRssBytes": 574849024
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.056834032293409,
"inferenceMs": 8.497625007294118,
"masksPerSecond": 117.67993988221751,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.056834032293409,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.53092771842765
},
"observedPeakRssBytes": 580304896
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.802125002257526,
"inferenceMs": 8.842708019074053,
"masksPerSecond": 113.08752905139042,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 13.802125002257526,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 45.67605298385659
},
"observedPeakRssBytes": 580304896
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 441.45958399167284,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.815792033914477,
"inferenceMs": 9.399125003255904,
"masksPerSecond": 106.39288227931803,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 456.2753760255873,
"processAndInitWallMs": 1996.0824169684201,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.97208515261655
},
"observedPeakRssBytes": 580304896
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.146541049238294,
"inferenceMs": 10.32191701233387,
"masksPerSecond": 96.88122843896919,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.146541049238294,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 39.13032816649965
},
"observedPeakRssBytes": 580304896
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.214833034202456,
"inferenceMs": 8.425000007264316,
"masksPerSecond": 118.69436191546191,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.214833034202456,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.94065277765507
},
"observedPeakRssBytes": 580304896
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.572583022527397,
"inferenceMs": 8.195583010092378,
"masksPerSecond": 122.01694483096064,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.572583022527397,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 49.282644017225
},
"observedPeakRssBytes": 580304896
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.7420420544222,
"inferenceMs": 9.763000009115785,
"masksPerSecond": 102.42753242510423,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.7420420544222,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 41.3704803464996
},
"observedPeakRssBytes": 580304896
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.073458023834974,
"inferenceMs": 9.47816704865545,
"masksPerSecond": 105.50563150729208,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4727029760,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.073458023834974,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.613724565795266
},
"observedPeakRssBytes": 580304896
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 150.4745000274852,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 3200.1133750309236,
"inferenceMs": 3185.321290977299,
"masksPerSecond": 0.3139400734339068,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 3350.587875058409,
"processAndInitWallMs": 1714.651208021678,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.21956717583908691
},
"observedPeakRssBytes": 2103443456,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.194917004555464,
"inferenceMs": 29.4447080232203,
"masksPerSecond": 33.96196013257775,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.194917004555464,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.752723221043816
},
"observedPeakRssBytes": 2103443456,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.30008299741894,
"inferenceMs": 29.568541038315743,
"masksPerSecond": 33.81972748348226,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.30008299741894,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.653246844127626
},
"observedPeakRssBytes": 2427781120,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.522332954220474,
"inferenceMs": 28.771542012691498,
"masksPerSecond": 34.756566038722816,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.522332954220474,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 24.308464234954428
},
"observedPeakRssBytes": 2427781120,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.1268340065144,
"inferenceMs": 29.463375045452267,
"masksPerSecond": 33.940442955273454,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.1268340065144,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.73767427937461
},
"observedPeakRssBytes": 2427781120,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.46975000528619,
"inferenceMs": 29.900832974817604,
"masksPerSecond": 33.44388435072017,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.46975000528619,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.39038516381888
},
"observedPeakRssBytes": 2427781120,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 150.4745000274852,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.7611250304617,
"inferenceMs": 30.848207999952137,
"masksPerSecond": 32.41679387021611,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 192.2356250579469,
"processAndInitWallMs": 1714.651208021678,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.672046298478186
},
"observedPeakRssBytes": 2440216576,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.0836249939166,
"inferenceMs": 30.490415985696018,
"masksPerSecond": 32.797191106514596,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 41.0836249939166,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.93809308236746
},
"observedPeakRssBytes": 2440216576,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.84170801797882,
"inferenceMs": 29.84916599234566,
"masksPerSecond": 33.50177355898098,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.84170801797882,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.430872412962824
},
"observedPeakRssBytes": 2440216576,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.927458019927144,
"inferenceMs": 30.86829202948138,
"masksPerSecond": 32.39570232926817,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 41.927458019927144,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.657295043471525
},
"observedPeakRssBytes": 2440216576,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.75933402823284,
"inferenceMs": 31.56491700792685,
"masksPerSecond": 31.680742254094046,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 42.75933402823284,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.157257686575342
},
"observedPeakRssBytes": 2440216576,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 43.13237505266443,
"inferenceMs": 31.92087501520291,
"masksPerSecond": 31.32746202990148,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 43.13237505266443,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 21.910176324016856
},
"observedPeakRssBytes": 2450866176,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 150.4745000274852,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 3040.2795000118203,
"inferenceMs": 3030.4313329979777,
"masksPerSecond": 0.329986028428075,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 3190.7540000393055,
"processAndInitWallMs": 1714.651208021678,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.1332813568820995
},
"observedPeakRssBytes": 2545008640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.970541002228856,
"inferenceMs": 32.18695800751448,
"masksPerSecond": 31.068484314874883,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 41.970541002228856,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 12.548560814777964
},
"observedPeakRssBytes": 2545008640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.411457990761846,
"inferenceMs": 31.57674998510629,
"masksPerSecond": 31.66887030716166,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 41.411457990761846,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 12.791056717062595
},
"observedPeakRssBytes": 2545008640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.0004999935627,
"inferenceMs": 29.242166958283633,
"masksPerSecond": 34.19719206947223,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.0004999935627,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.812245876859834
},
"observedPeakRssBytes": 2545008640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.42604101030156,
"inferenceMs": 29.379207990132272,
"masksPerSecond": 34.037677269444245,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.42604101030156,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.74781784912853
},
"observedPeakRssBytes": 2545008640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.24495802493766,
"inferenceMs": 29.430249996948987,
"masksPerSecond": 33.978644425503326,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.24495802493766,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.723974483460793
},
"observedPeakRssBytes": 2549219328,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 150.4745000274852,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.076208009850234,
"inferenceMs": 30.093375011347234,
"masksPerSecond": 33.22990524070273,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 190.55070803733543,
"processAndInitWallMs": 1714.651208021678,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.421558726719832
},
"observedPeakRssBytes": 2549219328,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 38.9702909742482,
"inferenceMs": 29.176417039707303,
"masksPerSecond": 34.27425645304774,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 38.9702909742482,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.84337218138598
},
"observedPeakRssBytes": 2549219328,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.61545799393207,
"inferenceMs": 29.79704097378999,
"masksPerSecond": 33.5603793973911,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.61545799393207,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.555037238606264
},
"observedPeakRssBytes": 2549219328,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.67604105127975,
"inferenceMs": 29.877750028390437,
"masksPerSecond": 33.46972242052296,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.67604105127975,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.518420885649224
},
"observedPeakRssBytes": 2549219328,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.036374994087964,
"inferenceMs": 30.250500014517456,
"masksPerSecond": 33.0573048220721,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.036374994087964,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.35184541763492
},
"observedPeakRssBytes": 2549219328,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.988333999644965,
"inferenceMs": 30.209457967430353,
"masksPerSecond": 33.10221590463912,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.988333999644965,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.36998500388374
},
"observedPeakRssBytes": 2549235712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 149.0931250154972,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 3183.3279169513844,
"inferenceMs": 3168.317958014086,
"masksPerSecond": 0.3156248877959218,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 3332.4210419668816,
"processAndInitWallMs": 1736.8250419967808,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.22074552152536536
},
"observedPeakRssBytes": 2104360960,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.97666702000424,
"inferenceMs": 29.64395796880126,
"masksPerSecond": 33.733687014819296,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.97666702000424,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.593070828668495
},
"observedPeakRssBytes": 2104360960,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.498292015399784,
"inferenceMs": 29.85533297760412,
"masksPerSecond": 33.494853356689966,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.498292015399784,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.42603247884211
},
"observedPeakRssBytes": 2429730816,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.52633396349847,
"inferenceMs": 29.84245802508667,
"masksPerSecond": 33.50930406467735,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.52633396349847,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.436139188402823
},
"observedPeakRssBytes": 2429730816,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.9948339802213,
"inferenceMs": 29.979374958202243,
"masksPerSecond": 33.35626581255337,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.9948339802213,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.329105459173324
},
"observedPeakRssBytes": 2429730816,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.190417007077485,
"inferenceMs": 30.779708991758525,
"masksPerSecond": 32.48893614516488,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 42.190417007077485,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.722502028439155
},
"observedPeakRssBytes": 2429730816,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 149.0931250154972,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.40133299026638,
"inferenceMs": 30.60791699681431,
"masksPerSecond": 32.67128567109224,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 191.4944580057636,
"processAndInitWallMs": 1736.8250419967808,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.850035828076543
},
"observedPeakRssBytes": 2442739712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 43.89866697601974,
"inferenceMs": 33.04104204289615,
"masksPerSecond": 30.265389290741233,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 43.89866697601974,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 21.167371146830092
},
"observedPeakRssBytes": 2442739712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.59479198278859,
"inferenceMs": 30.17450001789257,
"masksPerSecond": 33.14056568980531,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 41.59479198278859,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.17824651892431
},
"observedPeakRssBytes": 2442739712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.626624995842576,
"inferenceMs": 30.755291983950883,
"masksPerSecond": 32.51472951457696,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 41.626624995842576,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.74054170465901
},
"observedPeakRssBytes": 2442739712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.47112502343953,
"inferenceMs": 31.018875015433878,
"masksPerSecond": 32.23843545268601,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 42.47112502343953,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.54730384812498
},
"observedPeakRssBytes": 2453372928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.94204202899709,
"inferenceMs": 28.527709015179425,
"masksPerSecond": 35.053638533255715,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.94204202899709,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 24.51623436105078
},
"observedPeakRssBytes": 2453372928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 149.0931250154972,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 3094.305709004402,
"inferenceMs": 3084.2962090391666,
"masksPerSecond": 0.32422307464156447,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 3243.3988340198994,
"processAndInitWallMs": 1736.8250419967808,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.13095369984772787
},
"observedPeakRssBytes": 2538356736,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.45404099067673,
"inferenceMs": 29.601624992210418,
"masksPerSecond": 33.78192921041149,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.45404099067673,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.644521208085202
},
"observedPeakRssBytes": 2538356736,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.862333971541375,
"inferenceMs": 29.558334033936262,
"masksPerSecond": 33.83140602078211,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.862333971541375,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.664504891793893
},
"observedPeakRssBytes": 2538356736,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.05866695661098,
"inferenceMs": 29.922458983492106,
"masksPerSecond": 33.4197132846498,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.05866695661098,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.498222195670055
},
"observedPeakRssBytes": 2538356736,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.872583001852036,
"inferenceMs": 29.898582957684994,
"masksPerSecond": 33.4464011694228,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.872583001852036,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.50900143232987
},
"observedPeakRssBytes": 2548891648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.63379102060571,
"inferenceMs": 29.738834011368454,
"masksPerSecond": 33.62606616041919,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.63379102060571,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.58156812219331
},
"observedPeakRssBytes": 2548891648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 149.0931250154972,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.457457991782576,
"inferenceMs": 31.216625007800758,
"masksPerSecond": 32.03421253098658,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 190.55058300727978,
"processAndInitWallMs": 1736.8250419967808,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 12.938618441265477
},
"observedPeakRssBytes": 2548891648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.71404202841222,
"inferenceMs": 30.398416973184794,
"masksPerSecond": 32.89644986717976,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.71404202841222,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.286876101353906
},
"observedPeakRssBytes": 2548891648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.72870798315853,
"inferenceMs": 29.765124956611544,
"masksPerSecond": 33.59636492229394,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.72870798315853,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.569571792114521
},
"observedPeakRssBytes": 2548891648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.98904097918421,
"inferenceMs": 30.931874993257225,
"masksPerSecond": 32.3291103503421,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.98904097918421,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.057727670503173
},
"observedPeakRssBytes": 2548891648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.552500031888485,
"inferenceMs": 32.71066700108349,
"masksPerSecond": 30.57106722913588,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 42.552500031888485,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 12.34765405384798
},
"observedPeakRssBytes": 2548891648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.816958993673325,
"inferenceMs": 30.882957973517478,
"masksPerSecond": 32.380318001193814,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.816958993673325,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.07841044068218
},
"observedPeakRssBytes": 2548891648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 161.77825001068413,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 3191.5143330115825,
"inferenceMs": 3175.6790000363253,
"masksPerSecond": 0.31489328738470146,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 3353.2925830222666,
"processAndInitWallMs": 1772.176957980264,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.22023384605056115
},
"observedPeakRssBytes": 2102738944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.29070898145437,
"inferenceMs": 31.1670420342125,
"masksPerSecond": 32.08517506737681,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 42.29070898145437,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.440114760722803
},
"observedPeakRssBytes": 2430320640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 44.77708297781646,
"inferenceMs": 33.426040958147496,
"masksPerSecond": 29.916794551053556,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 44.77708297781646,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 20.92356677465045
},
"observedPeakRssBytes": 2430320640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.18016600841656,
"inferenceMs": 29.98320898041129,
"masksPerSecond": 33.352000469773685,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 41.18016600841656,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.32612231255596
},
"observedPeakRssBytes": 2430320640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 43.14229096053168,
"inferenceMs": 31.596000015269965,
"masksPerSecond": 31.64957588038714,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 43.14229096053168,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.135460174135723
},
"observedPeakRssBytes": 2430320640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.06075001275167,
"inferenceMs": 30.278332997113466,
"masksPerSecond": 33.026917304044886,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 42.06075001275167,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.09876174711056
},
"observedPeakRssBytes": 2430320640,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 161.77825001068413,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.57337499782443,
"inferenceMs": 30.450333026237786,
"masksPerSecond": 32.840363326678286,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 203.35162500850856,
"processAndInitWallMs": 1772.176957980264,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.96828738777218
},
"observedPeakRssBytes": 2437545984,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.85954197216779,
"inferenceMs": 30.085999984294176,
"masksPerSecond": 33.238050938045305,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.85954197216779,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.24642692166138
},
"observedPeakRssBytes": 2437545984,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.254499974660575,
"inferenceMs": 30.7412080001086,
"masksPerSecond": 32.52962603149711,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 42.254499974660575,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.75096020942083
},
"observedPeakRssBytes": 2437545984,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.238582973368466,
"inferenceMs": 31.24641696922481,
"masksPerSecond": 32.003669444241204,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 42.238582973368466,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.383110379946746
},
"observedPeakRssBytes": 2437545984,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 41.93954198854044,
"inferenceMs": 31.11816698219627,
"masksPerSecond": 32.13556892898393,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 41.93954198854044,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.475359824379925
},
"observedPeakRssBytes": 2453159936,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 45.42379203485325,
"inferenceMs": 34.07108300598338,
"masksPerSecond": 29.35040250479814,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 45.42379203485325,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 20.52743670863578
},
"observedPeakRssBytes": 2453159936,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 161.77825001068413,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 3059.906625014264,
"inferenceMs": 3049.9497079872526,
"masksPerSecond": 0.32787425883816557,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 3221.6848750249483,
"processAndInitWallMs": 1772.176957980264,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.13242841314473508
},
"observedPeakRssBytes": 2538160128,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.462000004481524,
"inferenceMs": 30.3471660008654,
"masksPerSecond": 32.952006127078995,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.462000004481524,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.309315274727206
},
"observedPeakRssBytes": 2538160128,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 38.36516698356718,
"inferenceMs": 28.430583013687283,
"masksPerSecond": 35.17339055335488,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 38.36516698356718,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 14.206532444500036
},
"observedPeakRssBytes": 2538160128,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.42687499895692,
"inferenceMs": 30.305709049571306,
"masksPerSecond": 32.997083102866576,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.42687499895692,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.32752186524781
},
"observedPeakRssBytes": 2538160128,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.48979201959446,
"inferenceMs": 29.118500009644777,
"masksPerSecond": 34.34242834173378,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.48979201959446,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.870906807226271
},
"observedPeakRssBytes": 2550792192,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.01937503926456,
"inferenceMs": 29.74074997473508,
"masksPerSecond": 33.62389989658987,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.01937503926456,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.580693168232647
},
"observedPeakRssBytes": 2550792192,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 161.77825001068413,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.30616697855294,
"inferenceMs": 30.371458968147635,
"masksPerSecond": 32.92564907891846,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 202.08441698923707,
"processAndInitWallMs": 1772.176957980264,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.298669662975165
},
"observedPeakRssBytes": 2550792192,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.78637500060722,
"inferenceMs": 30.75033298227936,
"masksPerSecond": 32.5199730544796,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.78637500060722,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.134817116704308
},
"observedPeakRssBytes": 2550792192,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.72591599449515,
"inferenceMs": 29.43445899290964,
"masksPerSecond": 33.97378563135427,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.72591599449515,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.722012016503989
},
"observedPeakRssBytes": 2550792192,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 40.63654103083536,
"inferenceMs": 30.296457989607006,
"masksPerSecond": 33.00715880196435,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 40.63654103083536,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.3315914401134
},
"observedPeakRssBytes": 2550792192,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.68524996889755,
"inferenceMs": 29.577541979961097,
"masksPerSecond": 33.80943557370332,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.68524996889755,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.65563102821877
},
"observedPeakRssBytes": 2550792192,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 39.88579200813547,
"inferenceMs": 29.99337500659749,
"masksPerSecond": 33.340696063048426,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 39.88579200813547,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 13.46630713986526
},
"observedPeakRssBytes": 2550792192,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 149.34412500588223,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1829.3096660054289,
"inferenceMs": 1821.4784999727271,
"masksPerSecond": 0.549004558667573,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1978.6537910113111,
"processAndInitWallMs": 1722.5562499952503,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.38396939629563126
},
"observedPeakRssBytes": 2428321792,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.546000017318875,
"inferenceMs": 13.801124994643033,
"masksPerSecond": 72.45786125320613,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.546000017318875,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 50.67644849760234
},
"observedPeakRssBytes": 2428321792,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.309250008314848,
"inferenceMs": 14.414540957659483,
"masksPerSecond": 69.37439096654882,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.309250008314848,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.519894046876516
},
"observedPeakRssBytes": 2428321792,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.61887496104464,
"inferenceMs": 14.030665974132717,
"masksPerSecond": 71.27245433991692,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.61887496104464,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 49.84738438570317
},
"observedPeakRssBytes": 2428321792,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.23687501018867,
"inferenceMs": 14.293499989435077,
"masksPerSecond": 69.96187083213641,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.23687501018867,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.93077276502955
},
"observedPeakRssBytes": 2428321792,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.5740829911083,
"inferenceMs": 14.811291999649256,
"masksPerSecond": 67.51605464423231,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.5740829911083,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.22018848973892
},
"observedPeakRssBytes": 2428321792,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 149.34412500588223,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.020082963630557,
"inferenceMs": 13.937207986600697,
"masksPerSecond": 71.75038221151648,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 167.3642079695128,
"processAndInitWallMs": 1722.5562499952503,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 50.18164331567693
},
"observedPeakRssBytes": 2434514944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.941416997928172,
"inferenceMs": 14.059708977583796,
"masksPerSecond": 71.12522752742305,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.941416997928172,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 49.744415130859466
},
"observedPeakRssBytes": 2434514944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.19691702257842,
"inferenceMs": 14.441333012655377,
"masksPerSecond": 69.24568522335645,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.19691702257842,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.42987827973371
},
"observedPeakRssBytes": 2434514944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.32841702364385,
"inferenceMs": 15.289667004253715,
"masksPerSecond": 65.40364807956847,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.32841702364385,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 45.74278823766556
},
"observedPeakRssBytes": 2434514944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.528542015701532,
"inferenceMs": 14.69904201803729,
"masksPerSecond": 68.03164442777248,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.528542015701532,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.58078785962864
},
"observedPeakRssBytes": 2434514944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.665957963094115,
"inferenceMs": 14.811291010119021,
"masksPerSecond": 67.51605915492469,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.665957963094115,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.22019164448108
},
"observedPeakRssBytes": 2451652608,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 149.34412500588223,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1773.0272079934366,
"inferenceMs": 1769.3281670217402,
"masksPerSecond": 0.5651862772767992,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1922.3713329993188,
"processAndInitWallMs": 1722.5562499952503,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.22827873739209914
},
"observedPeakRssBytes": 2534555648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.171457981225103,
"inferenceMs": 14.55491699744016,
"masksPerSecond": 68.70530420584842,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.171457981225103,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.750072368742174
},
"observedPeakRssBytes": 2534555648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.37274997867644,
"inferenceMs": 13.626666972413659,
"masksPerSecond": 73.38551694441773,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.37274997867644,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 29.64041029385032
},
"observedPeakRssBytes": 2534555648,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.900458013173193,
"inferenceMs": 14.37020901357755,
"masksPerSecond": 69.58841023503275,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.900458013173193,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.10675889392973
},
"observedPeakRssBytes": 2547154944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.98499998403713,
"inferenceMs": 14.323582989163697,
"masksPerSecond": 69.81493392795196,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.98499998403713,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.19825181349979
},
"observedPeakRssBytes": 2547154944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.67191698309034,
"inferenceMs": 14.062667032703757,
"masksPerSecond": 71.11026647181698,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.67191698309034,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.721436627966877
},
"observedPeakRssBytes": 2547154944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 149.34412500588223,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.281625001691282,
"inferenceMs": 14.587750018108636,
"masksPerSecond": 68.5506674270289,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 167.62575000757352,
"processAndInitWallMs": 1722.5562499952503,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.687614573776973
},
"observedPeakRssBytes": 2547154944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.53662497503683,
"inferenceMs": 13.82979197660461,
"masksPerSecond": 72.30766751167812,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.53662497503683,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 29.205066907966792
},
"observedPeakRssBytes": 2547154944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.470916965976357,
"inferenceMs": 13.894208997953683,
"masksPerSecond": 71.97243111481038,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.470916965976357,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 29.069664927271912
},
"observedPeakRssBytes": 2547154944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.107540963683277,
"inferenceMs": 14.370083983521909,
"masksPerSecond": 69.58901570420146,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.107540963683277,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.10700344292697
},
"observedPeakRssBytes": 2547154944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.44779199361801,
"inferenceMs": 13.909708999563009,
"masksPerSecond": 71.89223009851725,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.44779199361801,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 29.037271736791116
},
"observedPeakRssBytes": 2547154944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.391374964267015,
"inferenceMs": 14.65520798228681,
"masksPerSecond": 68.23512850917311,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.391374964267015,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.560168404855016
},
"observedPeakRssBytes": 2547154944,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 152.19804097432643,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1810.5068750446662,
"inferenceMs": 1803.2287080422975,
"masksPerSecond": 0.5545608250024286,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1962.7049160189927,
"processAndInitWallMs": 1732.5859580305405,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.3878554045200986
},
"observedPeakRssBytes": 2430681088,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.70854101050645,
"inferenceMs": 13.652208028361201,
"masksPerSecond": 73.24822460385839,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.70854101050645,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 51.22922230214173
},
"observedPeakRssBytes": 2430681088,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.834291957318783,
"inferenceMs": 14.931624988093972,
"masksPerSecond": 66.97194717904915,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.834291957318783,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 46.83964408144955
},
"observedPeakRssBytes": 2430681088,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.15387501846999,
"inferenceMs": 14.492291025817394,
"masksPerSecond": 69.00220249638535,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.15387501846999,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.259588408351945
},
"observedPeakRssBytes": 2430681088,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.26958399033174,
"inferenceMs": 14.408165996428579,
"masksPerSecond": 69.40508599414213,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.26958399033174,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.54136190361505
},
"observedPeakRssBytes": 2430681088,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.89725001435727,
"inferenceMs": 14.253208006266505,
"masksPerSecond": 70.15964402963489,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.89725001435727,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 49.06909375717441
},
"observedPeakRssBytes": 2430681088,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 152.19804097432643,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.14183294866234,
"inferenceMs": 15.039624995552003,
"masksPerSecond": 66.49101957633597,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 171.33987392298877,
"processAndInitWallMs": 1732.5859580305405,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 46.50328716353277
},
"observedPeakRssBytes": 2442493952,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.921917013358325,
"inferenceMs": 14.91529104532674,
"masksPerSecond": 67.0452890903071,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.921917013358325,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 46.89093882744806
},
"observedPeakRssBytes": 2442493952,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.144834029953927,
"inferenceMs": 14.369332988280803,
"masksPerSecond": 69.59265268718944,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.144834029953927,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.6725445481988
},
"observedPeakRssBytes": 2442493952,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.450125004164875,
"inferenceMs": 14.735458011273295,
"masksPerSecond": 67.86351664366012,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.450125004164875,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.463200632442735
},
"observedPeakRssBytes": 2442493952,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.136041006073356,
"inferenceMs": 14.502124977298081,
"masksPerSecond": 68.95541181484921,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.136041006073356,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.226863380011025
},
"observedPeakRssBytes": 2442493952,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.846291000954807,
"inferenceMs": 15.038000012282282,
"masksPerSecond": 66.4982044941648,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.846291000954807,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 46.50831223758291
},
"observedPeakRssBytes": 2458173440,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 152.19804097432643,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1713.9843340264633,
"inferenceMs": 1710.5374580132775,
"masksPerSecond": 0.5846115765050015,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1866.1823750007898,
"processAndInitWallMs": 1732.5859580305405,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.2361246157503701
},
"observedPeakRssBytes": 2541092864,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.690375039819628,
"inferenceMs": 14.158665959257632,
"masksPerSecond": 70.62812293739798,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.690375039819628,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.526698854415045
},
"observedPeakRssBytes": 2541092864,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.97724998323247,
"inferenceMs": 14.22370900399983,
"masksPerSecond": 70.30515034572146,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.97724998323247,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.396250224636894
},
"observedPeakRssBytes": 2541092864,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.22258300287649,
"inferenceMs": 13.638292031828314,
"masksPerSecond": 73.32296431739793,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.22258300287649,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 29.615145287797024
},
"observedPeakRssBytes": 2541092864,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.01954099209979,
"inferenceMs": 14.337625005282462,
"masksPerSecond": 69.74655841755984,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.01954099209979,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.17063494485242
},
"observedPeakRssBytes": 2553724928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.56154197594151,
"inferenceMs": 14.866417041048408,
"masksPerSecond": 67.26570344682582,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.56154197594151,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.16861762217295
},
"observedPeakRssBytes": 2553724928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 152.19804097432643,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.075084022711962,
"inferenceMs": 14.403791050426662,
"masksPerSecond": 69.42616679866225,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 170.2731249970384,
"processAndInitWallMs": 1732.5859580305405,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.041228769979682
},
"observedPeakRssBytes": 2553724928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.33658298710361,
"inferenceMs": 14.32608300819993,
"masksPerSecond": 69.80275064912176,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.33658298710361,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.193330987180282
},
"observedPeakRssBytes": 2553724928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.00375001039356,
"inferenceMs": 14.3494579824619,
"masksPerSecond": 69.68904339259458,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.00375001039356,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.14740462626895
},
"observedPeakRssBytes": 2553724928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.01958301803097,
"inferenceMs": 14.360582979861647,
"masksPerSecond": 69.6350560003264,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.01958301803097,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.125599118531834
},
"observedPeakRssBytes": 2553724928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.942874972708523,
"inferenceMs": 15.194542007520795,
"masksPerSecond": 65.81310575238354,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.942874972708523,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 26.581913413387706
},
"observedPeakRssBytes": 2553724928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.548916035797447,
"inferenceMs": 14.832749962806702,
"masksPerSecond": 67.41838179080158,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.548916035797447,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.230284405304754
},
"observedPeakRssBytes": 2553724928,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 159.20058300253004,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1798.4246669802815,
"inferenceMs": 1790.8892919658683,
"masksPerSecond": 0.5583818075668401,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1957.6252499828115,
"processAndInitWallMs": 1795.9323749528266,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.3905277691577874
},
"observedPeakRssBytes": 2393636864,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.676666950341314,
"inferenceMs": 14.919458015356213,
"masksPerSecond": 67.02656349652419,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.676666950341314,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 46.87784229696105
},
"observedPeakRssBytes": 2393636864,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.328959005884826,
"inferenceMs": 14.369666983839124,
"masksPerSecond": 69.59103513843793,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.328959005884826,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.67141324754239
},
"observedPeakRssBytes": 2393636864,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.757708021439612,
"inferenceMs": 14.677333005238324,
"masksPerSecond": 68.13226896487946,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.757708021439612,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.651163855884974
},
"observedPeakRssBytes": 2393636864,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.18949998030439,
"inferenceMs": 13.623957987874746,
"masksPerSecond": 73.40010890300711,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.18949998030439,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 51.33544896589195
},
"observedPeakRssBytes": 2393636864,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558808221998536,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.051959039643407,
"inferenceMs": 15.124124998692423,
"masksPerSecond": 66.11952758169193,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.051959039643407,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 46.243468634414675
},
"observedPeakRssBytes": 2431385600,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 159.20058300253004,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.640459038782865,
"inferenceMs": 15.65454195952043,
"masksPerSecond": 63.87922448231341,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 178.8410420413129,
"processAndInitWallMs": 1795.9323749528266,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 44.67661856913414
},
"observedPeakRssBytes": 2431385600,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.492250004783273,
"inferenceMs": 14.843542012386024,
"masksPerSecond": 67.36936501850847,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.492250004783273,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.117594939024684
},
"observedPeakRssBytes": 2431385600,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.441665975842625,
"inferenceMs": 14.642749971244484,
"masksPerSecond": 68.2931827671582,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.441665975842625,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.763705681888304
},
"observedPeakRssBytes": 2431385600,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.98133400734514,
"inferenceMs": 14.23566701123491,
"masksPerSecond": 70.24609378758238,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.98133400734514,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 49.12955602628482
},
"observedPeakRssBytes": 2431385600,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.41866597533226,
"inferenceMs": 14.609291043598205,
"masksPerSecond": 68.44959122353855,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.41866597533226,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.87309650501307
},
"observedPeakRssBytes": 2447491072,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23846140647877012,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.276042013894767,
"inferenceMs": 14.428417023736984,
"masksPerSecond": 69.30767237700746,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.276042013894767,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.473231599100004
},
"observedPeakRssBytes": 2447491072,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 159.20058300253004,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1696.453707991168,
"inferenceMs": 1692.6247919909656,
"masksPerSecond": 0.5907983888288324,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1855.654290993698,
"processAndInitWallMs": 1795.9323749528266,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.23862346924796538
},
"observedPeakRssBytes": 2532835328,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.937124997843057,
"inferenceMs": 14.297750021796674,
"masksPerSecond": 69.94107453798794,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.937124997843057,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.249200005893332
},
"observedPeakRssBytes": 2532835328,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.950792040210217,
"inferenceMs": 14.238707954064012,
"masksPerSecond": 70.23109141827577,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.950792040210217,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.366337823841583
},
"observedPeakRssBytes": 2532835328,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.151917029172182,
"inferenceMs": 14.359665976371616,
"masksPerSecond": 69.63950287182647,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.151917029172182,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.12739520993071
},
"observedPeakRssBytes": 2546499584,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.251209054142237,
"inferenceMs": 14.541333017405123,
"masksPerSecond": 68.76948618142907,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.251209054142237,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.775995468679206
},
"observedPeakRssBytes": 2546499584,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723124535776182,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.875834018923342,
"inferenceMs": 14.210916997399181,
"masksPerSecond": 70.36843577251317,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.875834018923342,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.42181120851807
},
"observedPeakRssBytes": 2546499584,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 159.20058300253004,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.16916698589921,
"inferenceMs": 14.458750025369227,
"masksPerSecond": 69.1622718592829,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 177.36974998842925,
"processAndInitWallMs": 1795.9323749528266,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.934641603964362
},
"observedPeakRssBytes": 2546499584,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.434791010804474,
"inferenceMs": 14.671374985482544,
"masksPerSecond": 68.15993736030255,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.434791010804474,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.529798699826202
},
"observedPeakRssBytes": 2546499584,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.406457966193557,
"inferenceMs": 14.618750021327287,
"masksPerSecond": 68.40530131106289,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.406457966193557,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.628901199538298
},
"observedPeakRssBytes": 2546499584,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.69291695766151,
"inferenceMs": 14.961750013753772,
"masksPerSecond": 66.83710121347688,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.69291695766151,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 26.995505180123313
},
"observedPeakRssBytes": 2546499584,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.02412496181205,
"inferenceMs": 14.349875040352345,
"masksPerSecond": 69.68701798363857,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.02412496181205,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.146586563591615
},
"observedPeakRssBytes": 2546565120,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733696459519683,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 15000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 739
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.14037503208965,
"inferenceMs": 14.475499978289008,
"masksPerSecond": 69.08224251320121,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.14037503208965,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.902317751081963
},
"observedPeakRssBytes": 2546565120,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 380.3788340301253,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 5025.739791977685,
"inferenceMs": 4992.414833046496,
"masksPerSecond": 0.20030386765551994,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3462725632,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 5406.11862600781,
"processAndInitWallMs": 1972.7248750277795,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.14009092260732942
},
"observedPeakRssBytes": 875954176
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 29.72545800730586,
"inferenceMs": 22.19708397751674,
"masksPerSecond": 45.05096259548743,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 29.72545800730586,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 31.508282831583145
},
"observedPeakRssBytes": 876707840
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 28.20579201215878,
"inferenceMs": 21.94258401868865,
"masksPerSecond": 45.57348392278198,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 28.20579201215878,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 31.873730067722338
},
"observedPeakRssBytes": 876707840
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.73712502978742,
"inferenceMs": 19.266666960902512,
"masksPerSecond": 51.903113394199494,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.73712502978742,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 36.300622282995974
},
"observedPeakRssBytes": 876707840
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 32.85837499424815,
"inferenceMs": 20.090000005438924,
"masksPerSecond": 49.77600795068552,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 32.85837499424815,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 34.81294175264585
},
"observedPeakRssBytes": 876707840
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.89129197783768,
"inferenceMs": 16.979917010758072,
"masksPerSecond": 58.89310291483897,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.89129197783768,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 41.18936503381506
},
"observedPeakRssBytes": 876707840
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 380.3788340301253,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 799.0846250322647,
"inferenceMs": 791.214041004423,
"masksPerSecond": 1.2638805028415943,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1179.46345906239,
"processAndInitWallMs": 1972.7248750277795,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.8839479126433883
},
"observedPeakRssBytes": 889257984
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 30.776667001191527,
"inferenceMs": 23.988833010662347,
"masksPerSecond": 41.686062825796014,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 30.776667001191527,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 29.154898851859127
},
"observedPeakRssBytes": 889257984
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.600999953690916,
"inferenceMs": 16.47012500325218,
"masksPerSecond": 60.715993339609795,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.600999953690916,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.464280013776374
},
"observedPeakRssBytes": 889257984
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.678208018187433,
"inferenceMs": 21.28454198827967,
"masksPerSecond": 46.98245330111636,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 27.678208018187433,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 32.85915197917438
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.725249994546175,
"inferenceMs": 22.00950001133606,
"masksPerSecond": 45.43492580408219,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 27.725249994546175,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 31.776823627968653
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.635208966676146,
"inferenceMs": 16.479166981298476,
"masksPerSecond": 60.68267899311043,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.635208966676146,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.44098022634949
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 380.3788340301253,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 613.6624590144493,
"inferenceMs": 606.9476249977015,
"masksPerSecond": 1.6475886201940193,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 994.0412930445746,
"processAndInitWallMs": 1972.7248750277795,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.6654610436963643
},
"observedPeakRssBytes": 891207680
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.85437502944842,
"inferenceMs": 9.608708031009883,
"masksPerSecond": 104.07226411425253,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.85437502944842,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.0347874757466
},
"observedPeakRssBytes": 891207680
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.083999991882592,
"inferenceMs": 11.632374953478575,
"masksPerSecond": 85.96696753666434,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.083999991882592,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 34.722058188058725
},
"observedPeakRssBytes": 891928576
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.19479100778699,
"inferenceMs": 20.42683301260695,
"masksPerSecond": 48.955214906922876,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 27.19479100778699,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 19.77301130090615
},
"observedPeakRssBytes": 891928576
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.35900001646951,
"inferenceMs": 14.917416032403708,
"masksPerSecond": 67.03573848364847,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 20.35900001646951,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.07573477354562
},
"observedPeakRssBytes": 891928576
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.742666981182992,
"inferenceMs": 19.47445800760761,
"masksPerSecond": 51.34931095948111,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.742666981182992,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 20.73998669653442
},
"observedPeakRssBytes": 891928576
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 380.3788340301253,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.545083007775247,
"inferenceMs": 14.472874987404794,
"masksPerSecond": 69.09477217693534,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 403.92391703790054,
"processAndInitWallMs": 1972.7248750277795,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.90737848226418
},
"observedPeakRssBytes": 891928576
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.382666975725442,
"inferenceMs": 18.01045797765255,
"masksPerSecond": 55.52329658917081,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.382666975725442,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.42585949236609
},
"observedPeakRssBytes": 891928576
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.725332994014025,
"inferenceMs": 16.777124954387546,
"masksPerSecond": 59.604968236138724,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.725332994014025,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 24.074446670576428
},
"observedPeakRssBytes": 891928576
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.119916022755206,
"inferenceMs": 17.00683298986405,
"masksPerSecond": 58.799895347710695,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.119916022755206,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.74927773094035
},
"observedPeakRssBytes": 891944960
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.270667006727308,
"inferenceMs": 16.893209016416222,
"masksPerSecond": 59.195384312609605,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.270667006727308,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.90901572386302
},
"observedPeakRssBytes": 891944960
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.012124991510063,
"inferenceMs": 17.713667009957135,
"masksPerSecond": 56.45358464951859,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.012124991510063,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.801602839940557
},
"observedPeakRssBytes": 891944960
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 352.10887499852106,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1116.980416001752,
"inferenceMs": 1107.0881669875234,
"masksPerSecond": 0.9032704258063574,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3462725632,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1469.089291000273,
"processAndInitWallMs": 1950.6389999878593,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.6317401096455599
},
"observedPeakRssBytes": 876216320
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.73575004376471,
"inferenceMs": 13.94295901991427,
"masksPerSecond": 71.72078742910547,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.73575004376471,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 50.160944961616934
},
"observedPeakRssBytes": 876216320
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.499125006608665,
"inferenceMs": 15.215291001368314,
"masksPerSecond": 65.72335684608791,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 20.499125006608665,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 45.96638999129912
},
"observedPeakRssBytes": 876216320
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.86612498573959,
"inferenceMs": 16.230084002017975,
"masksPerSecond": 61.613975619329175,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.86612498573959,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 43.09232163635387
},
"observedPeakRssBytes": 876216320
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.803208965342492,
"inferenceMs": 16.841541044414043,
"masksPerSecond": 59.37698915810779,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.803208965342492,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 41.52779120126733
},
"observedPeakRssBytes": 876216320
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.586832991801202,
"inferenceMs": 21.61733398679644,
"masksPerSecond": 46.2591733379697,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 27.586832991801202,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 32.3532957591893
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 352.10887499852106,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 71.75470801303163,
"inferenceMs": 65.41483401088044,
"masksPerSecond": 15.287052472435688,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 423.8635830115527,
"processAndInitWallMs": 1950.6389999878593,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 10.691642202801741
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 30.696958012413234,
"inferenceMs": 17.89212500443682,
"masksPerSecond": 55.890510476090675,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 30.696958012413234,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 39.089375902894005
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.735916987992823,
"inferenceMs": 17.11433398304507,
"masksPerSecond": 58.43055306684361,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.735916987992823,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 40.865861370525884
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 29.179792036302388,
"inferenceMs": 21.76125004189089,
"masksPerSecond": 45.95324248721823,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 29.179792036302388,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 32.139330169620536
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 29.718208010308444,
"inferenceMs": 24.20962502947077,
"masksPerSecond": 41.30588552208817,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 29.718208010308444,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.889005887064286
},
"observedPeakRssBytes": 890880000
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 28.210082964506,
"inferenceMs": 20.749457995407283,
"masksPerSecond": 48.194029946292645,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 28.210082964506,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 33.70651899219751
},
"observedPeakRssBytes": 890880000
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 352.10887499852106,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 949.9301670002751,
"inferenceMs": 943.9235409954563,
"masksPerSecond": 1.0594078403272005,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1302.0390419987962,
"processAndInitWallMs": 1950.6389999878593,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.42789482670815626
},
"observedPeakRssBytes": 890880000
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.552666027564555,
"inferenceMs": 10.752915986813605,
"masksPerSecond": 92.99802967179403,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.552666027564555,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 37.56190418443761
},
"observedPeakRssBytes": 890880000
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.333583022467792,
"inferenceMs": 12.659374973736703,
"masksPerSecond": 78.9928414376391,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.333583022467792,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 31.905208656662428
},
"observedPeakRssBytes": 890880000
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.177042016759515,
"inferenceMs": 13.149750011507422,
"masksPerSecond": 76.04707307172336,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.177042016759515,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 30.715412813669065
},
"observedPeakRssBytes": 890880000
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.22474999446422,
"inferenceMs": 13.792707992251962,
"masksPerSecond": 72.50207867532241,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.22474999446422,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 29.283589576962722
},
"observedPeakRssBytes": 890880000
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.235333024989814,
"inferenceMs": 14.819499978329986,
"masksPerSecond": 67.47865997248648,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.235333024989814,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.25463076288729
},
"observedPeakRssBytes": 891355136
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 352.10887499852106,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.952500000130385,
"inferenceMs": 19.855583959724754,
"masksPerSecond": 50.363666061316,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 378.06137499865144,
"processAndInitWallMs": 1950.6389999878593,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 20.34188472216553
},
"observedPeakRssBytes": 891355136
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 29.000332986470312,
"inferenceMs": 23.018500010948628,
"masksPerSecond": 43.44331731104786,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 29.000332986470312,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 17.546755861932233
},
"observedPeakRssBytes": 891355136
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 34.62591703282669,
"inferenceMs": 28.133499959949404,
"masksPerSecond": 35.54481317374628,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 34.62591703282669,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 14.356550040876122
},
"observedPeakRssBytes": 891355136
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.361415988299996,
"inferenceMs": 17.892415984533727,
"masksPerSecond": 55.88960154203903,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.361415988299996,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.573810062829562
},
"observedPeakRssBytes": 891355136
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.034166999626905,
"inferenceMs": 16.95550000295043,
"masksPerSecond": 58.977912761404255,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.034166999626905,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.821178964331178
},
"observedPeakRssBytes": 891355136
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 31.683792010881007,
"inferenceMs": 19.508957979269326,
"masksPerSecond": 51.25850396841407,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 31.683792010881007,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 20.70330975284244
},
"observedPeakRssBytes": 891355136
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 363.81741700461134,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1113.6701249633916,
"inferenceMs": 1104.8233750043437,
"masksPerSecond": 0.9051220517452108,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3462725632,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1477.487541968003,
"processAndInitWallMs": 1972.1579580218531,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.6330351220141864
},
"observedPeakRssBytes": 868646912
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.334124993998557,
"inferenceMs": 12.517208000645041,
"masksPerSecond": 79.89002019847139,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.334124993998557,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 55.8744410066493
},
"observedPeakRssBytes": 868646912
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.07375004561618,
"inferenceMs": 15.389041975140572,
"masksPerSecond": 64.9813030346787,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.07375004561618,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 45.44740349203001
},
"observedPeakRssBytes": 868646912
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.12270800070837,
"inferenceMs": 18.70562497060746,
"masksPerSecond": 53.45985507414593,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.12270800070837,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 37.38939496001707
},
"observedPeakRssBytes": 868646912
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.373292002361268,
"inferenceMs": 17.861208994872868,
"masksPerSecond": 55.98725149496062,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.373292002361268,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 39.157035797563495
},
"observedPeakRssBytes": 868646912
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558865414531479,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.823290994390845,
"inferenceMs": 18.552249995991588,
"masksPerSecond": 53.90181785045268,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.823290994390845,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 37.698500190063804
},
"observedPeakRssBytes": 874741760
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 363.81741700461134,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 77.36133300932124,
"inferenceMs": 70.6584159634076,
"masksPerSecond": 14.152595785870396,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 441.1787500139326,
"processAndInitWallMs": 1972.1579580218531,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 9.898212271871468
},
"observedPeakRssBytes": 874741760
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 28.335832990705967,
"inferenceMs": 18.683208036236465,
"masksPerSecond": 53.523998558517334,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 28.335832990705967,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 37.43425639983855
},
"observedPeakRssBytes": 874741760
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.355541041586548,
"inferenceMs": 17.163333017379045,
"masksPerSecond": 58.26374160470066,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.355541041586548,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 40.7491947683948
},
"observedPeakRssBytes": 874741760
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 26.51904203230515,
"inferenceMs": 20.594375033397228,
"masksPerSecond": 48.556948117062674,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 26.51904203230515,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 33.9603410574887
},
"observedPeakRssBytes": 874741760
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 32.15637500397861,
"inferenceMs": 26.363415992818773,
"masksPerSecond": 37.93135154687062,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 32.15637500397861,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 26.528883821068938
},
"observedPeakRssBytes": 889651200
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2383741878660322,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.651499956846237,
"inferenceMs": 21.228750003501773,
"masksPerSecond": 47.10592945110031,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 27.651499956846237,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 32.94551021066395
},
"observedPeakRssBytes": 889651200
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 363.81741700461134,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 699.836415995378,
"inferenceMs": 694.0785420010798,
"masksPerSecond": 1.440759135294582,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1063.6538329999894,
"processAndInitWallMs": 1972.1579580218531,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.5819226147454817
},
"observedPeakRssBytes": 889651200
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.631708956789225,
"inferenceMs": 9.723333001602441,
"masksPerSecond": 102.84539260716426,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.631708956789225,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 41.53925407403364
},
"observedPeakRssBytes": 889651200
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.346917011309415,
"inferenceMs": 11.944750032853335,
"masksPerSecond": 83.7187883588655,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.346917011309415,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 33.814018618145774
},
"observedPeakRssBytes": 890241024
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.911250019911677,
"inferenceMs": 16.38091600034386,
"masksPerSecond": 61.0466472069699,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.911250019911677,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 24.65674080689514
},
"observedPeakRssBytes": 890241024
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 28.335540962871164,
"inferenceMs": 22.6541250012815,
"masksPerSecond": 44.14207125384149,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 28.335540962871164,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 17.82898257942658
},
"observedPeakRssBytes": 890241024
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3723248328794256,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.58624997921288,
"inferenceMs": 16.84225001372397,
"masksPerSecond": 59.374489702097186,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.58624997921288,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.981356390677053
},
"observedPeakRssBytes": 890241024
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 363.81741700461134,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.95350004453212,
"inferenceMs": 20.123958005569875,
"masksPerSecond": 49.69201385349849,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 391.77091704914346,
"processAndInitWallMs": 1972.1579580218531,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 20.070604395428038
},
"observedPeakRssBytes": 890241024
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.095582976471633,
"inferenceMs": 21.341249987017363,
"masksPerSecond": 46.85761146176233,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 27.095582976471633,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 18.925789269405804
},
"observedPeakRssBytes": 890241024
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 28.16191699821502,
"inferenceMs": 18.382917041890323,
"masksPerSecond": 54.39833067413819,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 28.16191699821502,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 21.971485759284413
},
"observedPeakRssBytes": 890241024
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.142874976154417,
"inferenceMs": 18.77274998696521,
"masksPerSecond": 53.268700680206486,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.142874976154417,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 21.5152282047354
},
"observedPeakRssBytes": 890257408
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 31.13854199182242,
"inferenceMs": 22.396249987650663,
"masksPerSecond": 44.650332111465175,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 31.13854199182242,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 18.034269139820786
},
"observedPeakRssBytes": 890257408
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3733597425105224,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 739
},
"parameterDtype": "torch.float16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.6564590013586,
"inferenceMs": 17.638625053223222,
"masksPerSecond": 56.69376138914316,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "fp16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.6564590013586,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.898610225074922
},
"observedPeakRssBytes": 890257408
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": false,
"previousWarmupUsabilityPassed": false,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 367.54291702527553,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 5351.096583006438,
"inferenceMs": 5322.414166003,
"masksPerSecond": 0.18788466451699962,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3462725632,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 5718.639500031713,
"processAndInitWallMs": 1996.653666952625,
"grade": "C",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.1314050312858734
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 35.21316696424037,
"inferenceMs": 28.894999995827675,
"masksPerSecond": 34.60806368383444,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 35.21316696424037,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 24.204602875964337
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 35.346666001714766,
"inferenceMs": 29.956458020024,
"masksPerSecond": 33.38178363181532,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 35.346666001714766,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.346952417822582
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.658999977167696,
"inferenceMs": 17.257665982469916,
"masksPerSecond": 57.945263340696556,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.658999977167696,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 40.52645361837644
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 30.50287504447624,
"inferenceMs": 20.4395410255529,
"masksPerSecond": 48.92477765277753,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 30.50287504447624,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 34.21759809213138
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 30.3292500320822,
"inferenceMs": 24.27212498150766,
"masksPerSecond": 41.199524176885035,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 30.3292500320822,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.814617613119978
},
"observedPeakRssBytes": 876249088
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 367.54291702527553,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 794.1402499563992,
"inferenceMs": 788.1174999638461,
"masksPerSecond": 1.2688463332509097,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1161.6831669816747,
"processAndInitWallMs": 1996.653666952625,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.8874209747050202
},
"observedPeakRssBytes": 890159104
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 31.95112501271069,
"inferenceMs": 25.948834023438394,
"masksPerSecond": 38.53737702036037,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 31.95112501271069,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 26.952733189023878
},
"observedPeakRssBytes": 890159104
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.276875043753535,
"inferenceMs": 20.79404197866097,
"masksPerSecond": 48.09069833687019,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 27.276875043753535,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 33.63424969122031
},
"observedPeakRssBytes": 890159104
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.34441598597914,
"inferenceMs": 18.220583035144955,
"masksPerSecond": 54.88298580079133,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.34441598597914,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 38.384721205187056
},
"observedPeakRssBytes": 890159104
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 28.190209006424993,
"inferenceMs": 19.593334000092,
"masksPerSecond": 51.03776621147297,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 28.190209006424993,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 35.69540538617451
},
"observedPeakRssBytes": 890159104
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.771416967269033,
"inferenceMs": 16.357041953597218,
"masksPerSecond": 61.13574831175886,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.771416967269033,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.75785328325765
},
"observedPeakRssBytes": 890159104
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 367.54291702527553,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 934.0264169732109,
"inferenceMs": 927.1788750193082,
"masksPerSecond": 1.0785405351035153,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1301.5693339984864,
"processAndInitWallMs": 1996.653666952625,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.4356225221283098
},
"observedPeakRssBytes": 891158528
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.032833013217896,
"inferenceMs": 10.513374989386648,
"masksPerSecond": 95.11693447722634,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.032833013217896,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 38.41772983535172
},
"observedPeakRssBytes": 891158528
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.03924998966977,
"inferenceMs": 10.503875033464283,
"masksPerSecond": 95.20296050877425,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.03924998966977,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 38.45247574949392
},
"observedPeakRssBytes": 891158528
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.509124977048486,
"inferenceMs": 13.748999976087362,
"masksPerSecond": 72.7325624946707,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.509124977048486,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 29.376681991597493
},
"observedPeakRssBytes": 891158528
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.34708295064047,
"inferenceMs": 12.87179096834734,
"masksPerSecond": 77.68926658761565,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.34708295064047,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 31.378694774737962
},
"observedPeakRssBytes": 891158528
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.47537498967722,
"inferenceMs": 15.633832954335958,
"masksPerSecond": 63.963840660242916,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.47537498967722,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 25.834995242672115
},
"observedPeakRssBytes": 891158528
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 367.54291702527553,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.28079199185595,
"inferenceMs": 11.870666989125311,
"masksPerSecond": 84.24126470029843,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 385.8237090171315,
"processAndInitWallMs": 1996.653666952625,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 34.02504681245053
},
"observedPeakRssBytes": 894205952
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.96529202349484,
"inferenceMs": 12.991290946956724,
"masksPerSecond": 76.97464432772595,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.96529202349484,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 31.090058843968514
},
"observedPeakRssBytes": 894205952
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.300208045635372,
"inferenceMs": 14.044167008250952,
"masksPerSecond": 71.20393821951133,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.300208045635372,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.759270646860625
},
"observedPeakRssBytes": 894205952
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.95150002464652,
"inferenceMs": 15.245667018461972,
"masksPerSecond": 65.59240725834002,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.95150002464652,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 26.49277329164353
},
"observedPeakRssBytes": 894205952
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.734375019557774,
"inferenceMs": 18.588542006909847,
"masksPerSecond": 53.79658069085105,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.734375019557774,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 21.72843894103474
},
"observedPeakRssBytes": 894205952
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.519458016380668,
"inferenceMs": 14.636166975833476,
"masksPerSecond": 68.32389939600657,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.519458016380668,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 27.596022966047048
},
"observedPeakRssBytes": 894205952
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 347.68850001273677,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1155.666459002532,
"inferenceMs": 1144.6962080080993,
"masksPerSecond": 0.8735942278869893,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3462725632,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1503.3549590152688,
"processAndInitWallMs": 1931.0655419831164,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.6109848142303372
},
"observedPeakRssBytes": 864681984
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.769791043363512,
"inferenceMs": 8.589165983721614,
"masksPerSecond": 116.4257393436363,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 14.769791043363512,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 81.42723069102448
},
"observedPeakRssBytes": 864681984
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.73270801641047,
"inferenceMs": 12.480332981795073,
"masksPerSecond": 80.1260672658886,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.73270801641047,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 56.03953043722436
},
"observedPeakRssBytes": 864681984
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.14179198583588,
"inferenceMs": 17.43591600097716,
"masksPerSecond": 57.35288010930754,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.14179198583588,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 40.11214552540882
},
"observedPeakRssBytes": 864681984
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.79229199141264,
"inferenceMs": 18.27591599430889,
"masksPerSecond": 54.716819682876604,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.79229199141264,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 38.26850595164644
},
"observedPeakRssBytes": 864681984
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 31.661875022109598,
"inferenceMs": 24.122999981045723,
"masksPerSecond": 41.4542138534069,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 31.661875022109598,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.99274553536196
},
"observedPeakRssBytes": 874446848
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 347.68850001273677,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 67.45770800625905,
"inferenceMs": 60.635457979515195,
"masksPerSecond": 16.492000445314282,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 415.1462080189958,
"processAndInitWallMs": 1931.0655419831164,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 11.534373175449247
},
"observedPeakRssBytes": 874446848
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 31.018500041682273,
"inferenceMs": 23.51716603152454,
"masksPerSecond": 42.522130373171215,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 31.018500041682273,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 29.739637805952963
},
"observedPeakRssBytes": 874446848
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.27599998190999,
"inferenceMs": 18.479041988030076,
"masksPerSecond": 54.11535947847062,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.27599998190999,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 37.84784949636653
},
"observedPeakRssBytes": 874446848
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 28.339541982859373,
"inferenceMs": 17.992125009186566,
"masksPerSecond": 55.57987172106751,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 28.339541982859373,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 38.872117642740854
},
"observedPeakRssBytes": 889798656
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.48983400547877,
"inferenceMs": 13.659208023454994,
"masksPerSecond": 73.2106867603776,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.48983400547877,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 51.20296863471401
},
"observedPeakRssBytes": 889798656
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.570750003680587,
"inferenceMs": 19.58354201633483,
"masksPerSecond": 51.0632856490358,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.570750003680587,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 35.71325347665044
},
"observedPeakRssBytes": 889798656
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 347.68850001273677,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1016.9796250411309,
"inferenceMs": 1009.2134999576956,
"masksPerSecond": 0.9908706136431173,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1364.6681250538677,
"processAndInitWallMs": 1931.0655419831164,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.40021264085045505
},
"observedPeakRssBytes": 889798656
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.012834025081247,
"inferenceMs": 10.974416974931955,
"masksPerSecond": 91.12101374352967,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.012834025081247,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 36.803777451011634
},
"observedPeakRssBytes": 889798656
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.033667023293674,
"inferenceMs": 10.231499967630953,
"masksPerSecond": 97.7373799700597,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.033667023293674,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 39.476127769907116
},
"observedPeakRssBytes": 889798656
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.553208977915347,
"inferenceMs": 11.786791030317545,
"masksPerSecond": 84.8407337864765,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.553208977915347,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 34.26717237635786
},
"observedPeakRssBytes": 889798656
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.204874999355525,
"inferenceMs": 12.615458050277084,
"masksPerSecond": 79.26783126023999,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.204874999355525,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 32.01627704601093
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.022916967514902,
"inferenceMs": 13.174167019315064,
"masksPerSecond": 75.90612738808217,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 19.022916967514902,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 30.658484852046385
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 347.68850001273677,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.4401249862276,
"inferenceMs": 12.148417008575052,
"masksPerSecond": 82.3152513857684,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 366.12862499896437,
"processAndInitWallMs": 1931.0655419831164,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 33.24713003471186
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.134832964278758,
"inferenceMs": 12.416624987963587,
"masksPerSecond": 80.53718308875229,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 18.134832964278758,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 32.52896824954704
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.092584007419646,
"inferenceMs": 13.383832992985845,
"masksPerSecond": 74.71701122720799,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.092584007419646,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 30.178200834669305
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.08820795547217,
"inferenceMs": 13.204541988670826,
"masksPerSecond": 75.7315172959407,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.08820795547217,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 30.58795983583045
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.856625019107014,
"inferenceMs": 18.147457973100245,
"masksPerSecond": 55.10413642959183,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.856625019107014,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.25656070391214
},
"observedPeakRssBytes": 889962496
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.004499951843172,
"inferenceMs": 17.923542007338256,
"masksPerSecond": 55.792543660766384,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 24.004499951843172,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.534608384583542
},
"observedPeakRssBytes": 889978880
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 374.4281670078635,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1136.7521659703925,
"inferenceMs": 1126.5378749812953,
"masksPerSecond": 0.8876754365818403,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3462725632,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1511.180332978256,
"processAndInitWallMs": 2054.230500012636,
"grade": "B",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.6208330989418465
},
"observedPeakRssBytes": 843857920
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.0030000153929,
"inferenceMs": 11.503499990794808,
"masksPerSecond": 86.93006483246037,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.0030000153929,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 60.79819190330413
},
"observedPeakRssBytes": 843857920
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.023749992717057,
"inferenceMs": 16.55908301472664,
"masksPerSecond": 60.38981742592032,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.023749992717057,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.236155189149265
},
"observedPeakRssBytes": 843857920
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.282500005327165,
"inferenceMs": 13.850833987817168,
"masksPerSecond": 72.19781862085517,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 21.282500005327165,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 50.49457676087714
},
"observedPeakRssBytes": 843857920
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.774500015657395,
"inferenceMs": 16.130291973240674,
"masksPerSecond": 61.99515803303181,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.774500015657395,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 43.35891756703818
},
"observedPeakRssBytes": 876167168
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.2558965501464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.211415973491967,
"inferenceMs": 15.956291987095028,
"masksPerSecond": 62.671202106903664,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3464822784,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.211415973491967,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 43.83173738395157
},
"observedPeakRssBytes": 876167168
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 374.4281670078635,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 62.39649996859953,
"inferenceMs": 55.07958302041516,
"masksPerSecond": 18.1555477576755,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 436.82466697646305,
"processAndInitWallMs": 2054.230500012636,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 12.697844857336184
},
"observedPeakRssBytes": 876167168
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.52291700663045,
"inferenceMs": 17.28508301312104,
"masksPerSecond": 57.85335246818913,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 23.52291700663045,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 40.46217188943173
},
"observedPeakRssBytes": 876167168
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 28.779666987247765,
"inferenceMs": 19.365208980161697,
"masksPerSecond": 51.63899863019449,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 28.779666987247765,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 36.11590252996899
},
"observedPeakRssBytes": 890765312
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.385749991983175,
"inferenceMs": 18.71012500487268,
"masksPerSecond": 53.44699726696481,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.385749991983175,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 37.38040231253705
},
"observedPeakRssBytes": 890765312
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.513332977425307,
"inferenceMs": 18.15254194661975,
"masksPerSecond": 55.08870344112956,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.513332977425307,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 38.528598477098484
},
"observedPeakRssBytes": 890765312
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23818688232064422,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.82366704195738,
"inferenceMs": 18.305375007912517,
"masksPerSecond": 54.62876338604093,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 27.82366704195738,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 38.20692008208994
},
"observedPeakRssBytes": 890765312
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 374.4281670078635,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 989.7577500087209,
"inferenceMs": 983.9151669875719,
"masksPerSecond": 1.0163477843945374,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 1364.1859170165844,
"processAndInitWallMs": 2054.230500012636,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 0.41050287011695363
},
"observedPeakRssBytes": 891289600
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.052792034111917,
"inferenceMs": 8.299000037368387,
"masksPerSecond": 120.49644481229572,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.052792034111917,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 48.66851405968624
},
"observedPeakRssBytes": 891289600
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.971667016856372,
"inferenceMs": 8.444250037427992,
"masksPerSecond": 118.42377897002525,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.971667016856372,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 47.8313643259932
},
"observedPeakRssBytes": 891289600
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.692249988205731,
"inferenceMs": 9.415209002327174,
"masksPerSecond": 106.21113134640221,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 15.692249988205731,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 42.89867595081185
},
"observedPeakRssBytes": 892026880
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.876666981261224,
"inferenceMs": 10.630583041347563,
"masksPerSecond": 94.06821771774027,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 16.876666981261224,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 37.99415313619529
},
"observedPeakRssBytes": 892026880
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3722555087893043,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.4846249865368,
"inferenceMs": 12.421166989952326,
"masksPerSecond": 80.50773335620683,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 17.4846249865368,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 32.51707350257193
},
"observedPeakRssBytes": 892026880
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000,
"observedUsabilityTargetMet": true,
"previousWarmupUsabilityPassed": true,
"note": "Frozenplan excludes initialwarmup from warm-speed selection; original measurement and prior target score retained."
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 374.4281670078635,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.88395901955664,
"inferenceMs": 14.043874980416149,
"masksPerSecond": 71.2054188316598,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 395.31212602742016,
"processAndInitWallMs": 2054.230500012636,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 28.759868666107394
},
"observedPeakRssBytes": 892026880
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.164332997519523,
"inferenceMs": 17.773292027413845,
"masksPerSecond": 56.26419677669067,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.164332997519523,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 22.72510907810536
},
"observedPeakRssBytes": 892026880
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.726249997504056,
"inferenceMs": 16.954416991211474,
"masksPerSecond": 58.98168014378566,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.726249997504056,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.822700610075028
},
"observedPeakRssBytes": 892026880
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.350833020638674,
"inferenceMs": 17.197624954860657,
"masksPerSecond": 58.147564133113896,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.350833020638674,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 23.4858011533647
},
"observedPeakRssBytes": 892026880
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.056542013771832,
"inferenceMs": 19.40599997760728,
"masksPerSecond": 51.53045455807003,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 25.056542013771832,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 20.813150596004483
},
"observedPeakRssBytes": 892026880
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3732458529338945,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"usabilityPassed": true,
"maximumWarmInferenceMs": 5000
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 739
},
"parameterDtype": "torch.bfloat16",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.261542035266757,
"inferenceMs": 14.981666987296194,
"masksPerSecond": 66.74824642998385,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3470835712,
"memoryMeasurementScope": "500ms sampled helper RSS; not validated minimum or OS peak",
"precision": "bf16",
"cpuThreads": 4,
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"totalMs": 22.261542035266757,
"processAndInitWallMs": null,
"grade": "A",
"performance": {
"metric": "megapixelsPerSecond",
"value": 26.959616733070472
},
"observedPeakRssBytes": 892043264
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"nativeParity": true,
"visualizationEnvelopeValid": true,
"scope": "Exact native task result matches independently quality-passing matrix row; self-contained visual result added and validated."
},
"runtime": {
"acceleration": "partial",
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"cpuThreads": 4,
"device": "mps",
"driverAllocatedMemoryBytes": 4721016832,
"durationMs": 49.781125,
"engine": "transformers",
"fallback": false,
"fallbackReason": null,
"family": {
"catalogId": "sam21-native",
"defaultModelId": "sam21-small-native",
"displayName": "SAM 2.1",
"id": "sam-v2-1",
"variantId": "large",
"variantName": "Large"
},
"imageEmbeddingReused": true,
"inferenceMs": 8.331124961841851,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"loadMs": 2629.656292,
"masksPerSecond": 120.03180897900242,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelId": "sam21-large-native",
"parameterDtype": "torch.float32",
"placement": {
"mps:0/torch.float32": 739
},
"precision": "fp32",
"processedShape": [
1,
3,
1024,
1024
],
"provider": "Metal/MPS",
"requestedDevice": "mps",
"requestedPrecision": "fp32",
"reused": false,
"selectionReason": "Explicit device selection; actual placement is reported after inference",
"torchVersion": "2.10.0",
"version": "5.22.1",
"visualizationMs": 2.40975,
"totalMs": 2679.437417,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"nativeParity": true,
"visualizationEnvelopeValid": true,
"scope": "Exact native task result matches independently quality-passing matrix row; self-contained visual result added and validated."
},
"runtime": {
"acceleration": "partial",
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"cpuThreads": 4,
"device": "mps",
"driverAllocatedMemoryBytes": 4727029760,
"durationMs": 52.610125000000004,
"engine": "transformers",
"fallback": false,
"fallbackReason": null,
"family": {
"catalogId": "sam21-native",
"defaultModelId": "sam21-small-native",
"displayName": "SAM 2.1",
"id": "sam-v2-1",
"variantId": "large",
"variantName": "Large"
},
"imageEmbeddingReused": true,
"inferenceMs": 10.016000014729798,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"loadMs": 2629.656292,
"masksPerSecond": 99.84025544422656,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelId": "sam21-large-native",
"parameterDtype": "torch.float32",
"placement": {
"mps:0/torch.float32": 739
},
"precision": "fp32",
"processedShape": [
1,
3,
1024,
1024
],
"provider": "Metal/MPS",
"requestedDevice": "mps",
"requestedPrecision": "fp32",
"reused": false,
"selectionReason": "Explicit device selection; actual placement is reported after inference",
"torchVersion": "2.10.0",
"version": "5.22.1",
"visualizationMs": 2.422709,
"totalMs": 2682.2664170000003,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"nativeParity": true,
"visualizationEnvelopeValid": true,
"scope": "Exact native task result matches independently quality-passing matrix row; self-contained visual result added and validated."
},
"runtime": {
"acceleration": "partial",
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"cpuThreads": 4,
"device": "mps",
"driverAllocatedMemoryBytes": 4727029760,
"durationMs": 39.756958000000004,
"engine": "transformers",
"fallback": false,
"fallbackReason": null,
"family": {
"catalogId": "sam21-native",
"defaultModelId": "sam21-small-native",
"displayName": "SAM 2.1",
"id": "sam-v2-1",
"variantId": "large",
"variantName": "Large"
},
"imageEmbeddingReused": true,
"inferenceMs": 7.981541974004358,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"loadMs": 2629.656292,
"masksPerSecond": 125.28907362223616,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelId": "sam21-large-native",
"parameterDtype": "torch.float32",
"placement": {
"mps:0/torch.float32": 739
},
"precision": "fp32",
"processedShape": [
1,
3,
1024,
1024
],
"provider": "Metal/MPS",
"requestedDevice": "mps",
"requestedPrecision": "fp32",
"reused": false,
"selectionReason": "Explicit device selection; actual placement is reported after inference",
"torchVersion": "2.10.0",
"version": "5.22.1",
"visualizationMs": 1.8663340000000002,
"totalMs": 2669.41325,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"nativeParity": true,
"visualizationEnvelopeValid": true,
"scope": "Exact native task result matches independently quality-passing matrix row; self-contained visual result added and validated."
},
"runtime": {
"acceleration": "partial",
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"artifactRevision": "665f8e2ad61cf5f53d65644ff27c8ee525124610",
"cpuThreads": 4,
"device": "mps",
"driverAllocatedMemoryBytes": 4727029760,
"durationMs": 41.207625,
"engine": "transformers",
"fallback": false,
"fallbackReason": null,
"family": {
"catalogId": "sam21-native",
"defaultModelId": "sam21-small-native",
"displayName": "SAM 2.1",
"id": "sam-v2-1",
"variantId": "large",
"variantName": "Large"
},
"imageEmbeddingReused": true,
"inferenceMs": 9.491708013229072,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"loadMs": 2629.656292,
"masksPerSecond": 105.35511612938888,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelId": "sam21-large-native",
"parameterDtype": "torch.float32",
"placement": {
"mps:0/torch.float32": 739
},
"precision": "fp32",
"processedShape": [
1,
3,
1024,
1024
],
"provider": "Metal/MPS",
"requestedDevice": "mps",
"requestedPrecision": "fp32",
"reused": false,
"selectionReason": "Explicit device selection; actual placement is reported after inference",
"torchVersion": "2.10.0",
"version": "5.22.1",
"visualizationMs": 1.960375,
"totalMs": 2670.863917,
"grade": "A"
}
}
]SAM2.1 Tiny — Object Selection Desktop
36 configurations testedQuality criteria
No explicit quality gate recorded. Review per-case evidence; full certification is pending.
| Precision | Device | Result | Platform | Inference / load | |
|---|---|---|---|---|---|
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 5346 ms / 181 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 886 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 124 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 117 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 271 ms / 185 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 262 ms / 174 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 0 ms |
FP32 | mpsAccelerated | Passed | Not measured | macos · aarch64 | 7 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 857 ms / 87 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 33 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 797 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 33 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 802 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 32 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 853 ms / 88 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 34 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 33 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 856 ms / 87 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 33 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 33 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 404 ms / 90 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 365 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 361 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 407 ms / 90 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 15 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 403 ms / 89 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
FP32 | cpu | Passed | Not measured | macos · aarch64 | 14 ms / 0 ms |
Test setup & sample details
- Quality gate
- No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
- Tested
- 2026-11-01T21:50:41.784719+00:00
- Environment
- { "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2", "python": "app-owned CPython3.13.18 private probe venv", "runtime": "Transformers5.22.1/Torch2.10.0" }
- Runtime
- transformers 5.22.1
- Artifact revision
- de431c4043854a71d8101e17995dfe596bf101a5
- Fixture
{ "file": "model-samples/modnet-portrait-pexels-5965592.jpg", "size": 78124, "sha1": "38212739a4a4f0be25d84863cf991d7fb6c6d818", "sha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe", "url": "https://assets.localai.tools/samples/v1/files/2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe/modnet-portrait-pexels-5965592.jpg" }- Rerun when
- artifact/runtime/processor/backend/driver; fixture checksum, prompt coordinates, expected probes/area gate; mask candidate selection/postprocessing and image-embedding cache semantics; Concrete variant source/artifact/dimension/architecture change
Per-case quality & failure details
[
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 87.0130419934867,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 871.7708750045858,
"inferenceMs": 857.3468329996103,
"masksPerSecond": 1.1663890989149481,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 958.7839169980725,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24879180774158127,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 44.10112500045216,
"inferenceMs": 33.32362500077579,
"masksPerSecond": 30.008740044839644,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 44.10112500045216,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3712205991582075,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 806.4343750011176,
"inferenceMs": 796.9382910014247,
"masksPerSecond": 1.2548022993642456,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 806.4343750011176,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3719064124783362,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.19741699489532,
"inferenceMs": 32.58804099459667,
"masksPerSecond": 30.68610353613484,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 42.19741699489532,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 812.4142500018934,
"inferenceMs": 801.861457999621,
"masksPerSecond": 1.2470982237937573,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 812.4142500018934,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 42.72833299910417,
"inferenceMs": 32.37224999611499,
"masksPerSecond": 30.890654808362427,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 42.72833299910417,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 87.9316659993492,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 867.71629100258,
"inferenceMs": 853.2166250006412,
"masksPerSecond": 1.1720352964280887,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 955.6479570019292,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 45.34141700423788,
"inferenceMs": 33.7382500001695,
"masksPerSecond": 29.6399487227398,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 45.34141700423788,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 43.81783299322706,
"inferenceMs": 32.97533399745589,
"masksPerSecond": 30.325697385723277,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 43.81783299322706,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 86.83362499868963,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 870.5124159969273,
"inferenceMs": 856.267208000645,
"masksPerSecond": 1.1678597412774525,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 957.3460409956169,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 44.11254199658288,
"inferenceMs": 33.181707993207965,
"masksPerSecond": 30.137086379178918,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 44.11254199658288,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 43.54641700047068,
"inferenceMs": 33.04937500070082,
"masksPerSecond": 30.25775827769193,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 1,
"totalMs": 43.54641700047068,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 90.23412499664119,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 412.13162500207545,
"inferenceMs": 403.9459589985199,
"masksPerSecond": 2.4755786701747002,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 502.36574999871664,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.24879180774158127,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.634125004406087,
"inferenceMs": 13.794790997053497,
"masksPerSecond": 72.49113090684703,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.634125004406087,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3712205991582075,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 369.07533399789827,
"inferenceMs": 365.46816599729937,
"masksPerSecond": 2.7362164287857276,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 369.07533399789827,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3719064124783362,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.292042000917718,
"inferenceMs": 13.761666996288113,
"masksPerSecond": 72.66561531170072,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.292042000917718,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 364.59291700157337,
"inferenceMs": 360.8993340021698,
"masksPerSecond": 2.7708557644331586,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 364.59291700157337,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.565333997481503,
"inferenceMs": 13.865458000509534,
"masksPerSecond": 72.12167098722968,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.565333997481503,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 90.17149999999674,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 414.1441660030978,
"inferenceMs": 407.3439159983536,
"masksPerSecond": 2.454928036789536,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 504.3156660030945,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.717459006235003,
"inferenceMs": 14.019208996614907,
"masksPerSecond": 71.33070062950492,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.717459006235003,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.336708002607338,
"inferenceMs": 14.576749999832828,
"masksPerSecond": 68.60239765458476,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 18.336708002607338,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 88.64195800560992,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 409.4957919951412,
"inferenceMs": 402.9149579946534,
"masksPerSecond": 2.4819133173340013,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 498.1377500007511,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.608624999411404,
"inferenceMs": 13.807415998599026,
"masksPerSecond": 72.42484763995417,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.608624999411404,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.25581076134699854,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.559291001816746,
"inferenceMs": 13.803667003230657,
"masksPerSecond": 72.44451780573644,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 17.559291001816746,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.255662060761347,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 181.36687499645632,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 5355.706458998611,
"inferenceMs": 5346.030332999362,
"masksPerSecond": 0.18705468127019684,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2203254784,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 5537.0733339950675,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "C"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.23864299277086384,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 894.5950839988654,
"inferenceMs": 886.1670839978615,
"masksPerSecond": 1.1284553647474602,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2213462016,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 894.5950839988654,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.36989106214409506,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 130.7745420053834,
"inferenceMs": 124.19204199977685,
"masksPerSecond": 8.052045718048479,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3287203840,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 130.7745420053834,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.3710745233968804,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.688875002320856,
"inferenceMs": 6.915999998454936,
"masksPerSecond": 144.59224988771027,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3287203840,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 14.688875002320856,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.255662060761347,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 124.44720899657113,
"inferenceMs": 117.14070799644105,
"masksPerSecond": 8.536741984096441,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3287203840,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 124.44720899657113,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.255662060761347,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.031583996256813,
"inferenceMs": 6.723040998622309,
"masksPerSecond": 148.74221356152978,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3287203840,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 14.031583996256813,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.255662060761347,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 184.67729199619498,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 280.79229200375266,
"inferenceMs": 271.39733300282387,
"masksPerSecond": 3.6846345869935098,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2203271168,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 465.46958399994764,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.255662060761347,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.927332998719066,
"inferenceMs": 7.217040998511948,
"masksPerSecond": 138.56094210995695,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2207465472,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 13.927332998719066,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.255662060761347,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 13.007665998884477,
"inferenceMs": 7.24866700329585,
"masksPerSecond": 137.95639936905866,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2207465472,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 13.007665998884477,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.255662060761347,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 174.4014590003644,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 271.6798330002348,
"inferenceMs": 261.8377920007333,
"masksPerSecond": 3.8191583894703762,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2203271168,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 446.0812920005992,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "B"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.255662060761347,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.735416996700224,
"inferenceMs": 7.010291999904439,
"masksPerSecond": 142.64741040938546,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2207465472,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 14.735416996700224,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independently annotated foreground/background probe accuracy plus whole-subject area bounds",
"score": 1,
"threshold": 1,
"foregroundRatio": 0.255662060761347,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 0,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 307
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 14.411249998374842,
"inferenceMs": 6.803167001635302,
"masksPerSecond": 146.9903648932367,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 2207465472,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"precision": "fp32",
"cpuThreads": 4,
"totalMs": 14.411249998374842,
"artifactRevision": "de431c4043854a71d8101e17995dfe596bf101a5",
"fallback": false,
"grade": "A"
}
}
]