SlimSAM Desktop
Select a whole object with a box or include/exclude points using the compact 77%-pruned SlimSAM Uniform checkpoint. Returns a complete binary mask, candidate scores and optional visualization.
Desktop app
Run SlimSAM 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.
SlimSAM 77 Uniform Desktop
125 configurations testedQuality criteria
{ "metric": "Independent foreground/background probe accuracy and whole-subject area bounds", "minimumProbeAccuracy": 0.9, "requirePromptSemantics": true, "frozenPolicy": "research/slimsam-family-expansion/quality-gates.json" }
| Precision | Device | Result | Platform | Inference / load | |
|---|---|---|---|---|---|
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 47 ms / 4792 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 48 ms / 4685 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 49 ms / 4576 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 49 ms / 4479 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 49 ms / 4536 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 49 ms / 4460 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 50 ms / 4388 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 50 ms / 4475 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 52 ms / 4455 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 63 ms / 4792 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 63 ms / 4536 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 64 ms / 4460 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 64 ms / 4536 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 64 ms / 4685 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 64 ms / 4479 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 64 ms / 4576 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 65 ms / 4388 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 65 ms / 4792 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4388 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4685 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4455 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4475 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4475 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4479 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 67 ms / 4460 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 67 ms / 4576 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 69 ms / 4455 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 83 ms / 4754 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 84 ms / 4716 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 88 ms / 6308 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 92 ms / 4388 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 96 ms / 4685 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 97 ms / 4536 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 98 ms / 4475 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 99 ms / 4460 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 100 ms / 4792 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 101 ms / 4479 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 105 ms / 4576 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 106 ms / 4455 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 387 ms / 4685 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 389 ms / 4536 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 389 ms / 4460 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 391 ms / 4576 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 393 ms / 4792 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 397 ms / 4455 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 401 ms / 4685 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 402 ms / 4536 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 402 ms / 4792 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 404 ms / 4460 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 405 ms / 4576 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 405 ms / 4388 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 406 ms / 4479 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 408 ms / 4455 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 408 ms / 4475 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 422 ms / 4475 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 422 ms / 4479 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 425 ms / 4388 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 513 ms / 7761 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 518 ms / 4617 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 542 ms / 4716 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 543 ms / 4685 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 545 ms / 4536 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 547 ms / 4490 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 548 ms / 4754 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 548 ms / 4460 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 566 ms / 4455 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 573 ms / 4792 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 577 ms / 4388 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 581 ms / 4475 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 592 ms / 4576 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 609 ms / 4479 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 668 ms / 6308 ms |
FP16 | mpsAccelerated | Passed | Moderate | macos · aarch64 | 2114 ms / 4490 ms |
BF16 | mpsAccelerated | Passed | Moderate | macos · aarch64 | 2238 ms / 4617 ms |
FP32 | mpsAccelerated | Passed | Moderate | macos · aarch64 | 3049 ms / 7761 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 47 ms / 4516 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 48 ms / 4469 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 48 ms / 4499 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 49 ms / 4416 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 52 ms / 4441 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 53 ms / 4499 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 54 ms / 4425 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 58 ms / 4338 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 62 ms / 4516 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 62 ms / 4516 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 62 ms / 4469 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 63 ms / 4416 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 63 ms / 4416 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 63 ms / 4516 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 63 ms / 4469 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 63 ms / 4416 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 64 ms / 4469 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 64 ms / 4729 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 67 ms / 4499 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 67 ms / 4425 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 68 ms / 4499 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 68 ms / 4425 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 69 ms / 4499 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 69 ms / 4338 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 69 ms / 4425 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 70 ms / 4338 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 71 ms / 4730 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 73 ms / 4338 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 463 ms / 4469 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 466 ms / 4416 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 469 ms / 4516 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 477 ms / 4469 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 483 ms / 4516 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 484 ms / 4416 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 539 ms / 4499 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 558 ms / 4416 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 560 ms / 4516 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 566 ms / 4469 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 578 ms / 4729 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 682 ms / 4499 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 684 ms / 4338 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 685 ms / 4425 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 703 ms / 4425 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 712 ms / 4499 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 713 ms / 4338 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 808 ms / 4441 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 819 ms / 4338 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 829 ms / 4499 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 831 ms / 4730 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 868 ms / 4425 ms |
Test setup & sample details
- Quality gate
- { "metric": "Independent foreground/background probe accuracy and whole-subject area bounds", "minimumProbeAccuracy": 0.9, "requirePromptSemantics": true, "frozenPolicy": "research/slimsam-family-expansion/quality-gates.json" }
- Tested
- 2026-11-13T10:43:24.851097+00:00
- Environment
- { "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max;Mac15,11;96GiB", "osVersion": "27.2", "runtime": "Transformers5.22.1/Torch2.10.0/Pillow12.0.0; MPS CPU-op fallback disabled", "transport": "Verified local original provider snapshots; no network-transfer timing", "concurrentUploadActivity": "none; uploads excluded" }
- Runtime
- transformers 5.22.1
- Artifact revision
- 79c09c1ce6b4ae51f00634ed171d9b8e888f6911
- 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": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4425.234583963174,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 789.5566250081174,
"inferenceMs": 868.0114999879152,
"masksPerSecond": 1.152058469287472,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 780.2123749861494,
"totalMs": 5293.2460839510895,
"residentMemoryBytes": 2246541312,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4425.234583963174,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.317291994113475,
"inferenceMs": 69.21891699312255,
"masksPerSecond": 14.44691774214494,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.395749989897013,
"totalMs": 4494.453500956297,
"residentMemoryBytes": 2248310784,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4425.234583963174,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 655.3919589496218,
"inferenceMs": 685.3099169675261,
"masksPerSecond": 1.4591938263858302,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 648.2017500093207,
"totalMs": 5110.5445009307,
"residentMemoryBytes": 2248491008,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4425.234583963174,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.92062499327585,
"inferenceMs": 53.67100000148639,
"masksPerSecond": 18.632035922049255,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 15.786333999130875,
"totalMs": 4478.905583964661,
"residentMemoryBytes": 2252701696,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4425.234583963174,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 657.9708330100402,
"inferenceMs": 703.0277079902589,
"masksPerSecond": 1.4224190435661406,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 652.2137499996461,
"totalMs": 5128.262291953433,
"residentMemoryBytes": 2252013568,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4425.234583963174,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.31429103994742,
"inferenceMs": 68.4479579795152,
"masksPerSecond": 14.609639637449456,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.460250015370548,
"totalMs": 4493.68254194269,
"residentMemoryBytes": 2251325440,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4425.234583963174,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.399416960775852,
"inferenceMs": 67.4082079785876,
"masksPerSecond": 14.834988645858273,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 15.47625003149733,
"totalMs": 4492.642791941762,
"residentMemoryBytes": 2250145792,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4499.072792008519,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 749.3477500393055,
"inferenceMs": 829.2162080178969,
"masksPerSecond": 1.2059580967312895,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 739.8820000234991,
"totalMs": 5328.289000026416,
"residentMemoryBytes": 2247524352,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4499.072792008519,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.70854201540351,
"inferenceMs": 67.75279197609052,
"masksPerSecond": 14.759539361166,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 15.823792025912553,
"totalMs": 4566.82558398461,
"residentMemoryBytes": 2249277440,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4499.072792008519,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 651.2499999953434,
"inferenceMs": 681.9194999989122,
"masksPerSecond": 1.4664487523844019,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 643.9745840034448,
"totalMs": 5180.992292007431,
"residentMemoryBytes": 2249474048,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4499.072792008519,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.152375011704862,
"inferenceMs": 53.31400001887232,
"masksPerSecond": 18.75679933312105,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 15.984083991497755,
"totalMs": 4552.3867920273915,
"residentMemoryBytes": 2251587584,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4499.072792008519,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 666.3003749563359,
"inferenceMs": 711.705332971178,
"masksPerSecond": 1.4050758841798605,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 660.5457500554621,
"totalMs": 5210.778124979697,
"residentMemoryBytes": 2250899456,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4499.072792008519,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.52966702124104,
"inferenceMs": 69.03629103908315,
"masksPerSecond": 14.48513506372866,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 15.575749974232167,
"totalMs": 4568.109083047602,
"residentMemoryBytes": 2250211328,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4499.072792008519,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.440208016429096,
"inferenceMs": 67.0131659717299,
"masksPerSecond": 14.922440769652024,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 15.590584021992981,
"totalMs": 4566.085957980249,
"residentMemoryBytes": 2249031680,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.266749982722,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 740.2098329621367,
"inferenceMs": 818.8305829535238,
"masksPerSecond": 1.2212538476433032,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 730.836917005945,
"totalMs": 5157.097332936246,
"residentMemoryBytes": 2245836800,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.266749982722,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.05395803321153,
"inferenceMs": 69.14179102750495,
"masksPerSecond": 14.463032923202626,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.12558402121067,
"totalMs": 4407.408541010227,
"residentMemoryBytes": 2247655424,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.266749982722,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 652.9772920184769,
"inferenceMs": 683.7671250104904,
"masksPerSecond": 1.4624862229003155,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 645.6917920149863,
"totalMs": 5022.0338749932125,
"residentMemoryBytes": 2247868416,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.266749982722,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 27.11500000441447,
"inferenceMs": 57.686207990627736,
"masksPerSecond": 17.335166148596034,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 19.80212499620393,
"totalMs": 4395.95295797335,
"residentMemoryBytes": 2252079104,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.266749982722,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 663.3326250012033,
"inferenceMs": 712.958250020165,
"masksPerSecond": 1.4026066743343197,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 657.3952090111561,
"totalMs": 5051.225000002887,
"residentMemoryBytes": 2251390976,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.266749982722,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.238041972741485,
"inferenceMs": 72.84658297430724,
"masksPerSecond": 13.727479851082334,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.235792019870132,
"totalMs": 4411.113332957029,
"residentMemoryBytes": 2250719232,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.266749982722,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.949749953113496,
"inferenceMs": 69.83458303147927,
"masksPerSecond": 14.319552814530745,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.793792019598186,
"totalMs": 4408.101333014201,
"residentMemoryBytes": 2249539584,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4416.359250026289,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 480.7410840294324,
"inferenceMs": 557.5763329979964,
"masksPerSecond": 1.7934764099888607,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 473.7108330009505,
"totalMs": 4973.935583024286,
"residentMemoryBytes": 2248343552,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4416.359250026289,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.256916966289282,
"inferenceMs": 63.04779200581834,
"masksPerSecond": 15.860983678979835,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 13.11487500788644,
"totalMs": 4479.407042032108,
"residentMemoryBytes": 2251735040,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4416.359250026289,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 435.12687500333413,
"inferenceMs": 465.93116596341133,
"masksPerSecond": 2.1462397732769998,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 430.29862496769056,
"totalMs": 4882.290415989701,
"residentMemoryBytes": 2254045184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4416.359250026289,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.80837500700727,
"inferenceMs": 48.600625013932586,
"masksPerSecond": 20.575867074000076,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.77295802719891,
"totalMs": 4464.959875040222,
"residentMemoryBytes": 2256207872,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4416.359250026289,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 438.4415000095032,
"inferenceMs": 483.965834020637,
"masksPerSecond": 2.066261561673295,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 434.44362498121336,
"totalMs": 4900.325084046926,
"residentMemoryBytes": 2255536128,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4416.359250026289,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.514458984602243,
"inferenceMs": 62.66895798034966,
"masksPerSecond": 15.95686336947804,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.576957989949733,
"totalMs": 4479.028208006639,
"residentMemoryBytes": 2254848000,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4416.359250026289,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.294249973725528,
"inferenceMs": 62.815167009830475,
"masksPerSecond": 15.919722060812186,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.41295802174136,
"totalMs": 4479.17441703612,
"residentMemoryBytes": 2253668352,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4468.856833002064,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 487.5206659780815,
"inferenceMs": 566.137166984845,
"masksPerSecond": 1.7663563855484676,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 479.5206669950858,
"totalMs": 5034.993999986909,
"residentMemoryBytes": 2246721536,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4468.856833002064,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.786124964710325,
"inferenceMs": 63.52649995824322,
"masksPerSecond": 15.74146223477309,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.803834048099816,
"totalMs": 4532.383332960308,
"residentMemoryBytes": 2249523200,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4468.856833002064,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 432.3586249956861,
"inferenceMs": 463.18258397514,
"masksPerSecond": 2.1589758220565396,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 427.58825002238154,
"totalMs": 4932.039416977204,
"residentMemoryBytes": 2252079104,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4468.856833002064,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.62650004820898,
"inferenceMs": 48.12162497546524,
"masksPerSecond": 20.780678136073938,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.77333399048075,
"totalMs": 4516.97845797753,
"residentMemoryBytes": 2256142336,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4468.856833002064,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 431.344166980125,
"inferenceMs": 477.264083980117,
"masksPerSecond": 2.095276040175821,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 427.4589169654064,
"totalMs": 4946.120916982181,
"residentMemoryBytes": 2255470592,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4468.856833002064,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.261959041003138,
"inferenceMs": 63.01375001203269,
"masksPerSecond": 15.86955227722595,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.373125005979091,
"totalMs": 4531.870583014097,
"residentMemoryBytes": 2254782464,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4468.856833002064,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.03166601853445,
"inferenceMs": 62.43637501029298,
"masksPerSecond": 16.01630459544047,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.152624956797808,
"totalMs": 4531.293208012357,
"residentMemoryBytes": 2253602816,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4516.144959023222,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 481.0787499882281,
"inferenceMs": 559.6131249913014,
"masksPerSecond": 1.7869487961268669,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 473.9424999570474,
"totalMs": 5075.758084014524,
"residentMemoryBytes": 2247737344,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4516.144959023222,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.9788749772124,
"inferenceMs": 62.83716601319611,
"masksPerSecond": 15.914148639198578,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 13.016499986406416,
"totalMs": 4578.982125036418,
"residentMemoryBytes": 2251800576,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4516.144959023222,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 438.9922079863027,
"inferenceMs": 468.7878330005333,
"masksPerSecond": 2.1331611650400117,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 434.2633340274915,
"totalMs": 4984.932792023756,
"residentMemoryBytes": 2254143488,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4516.144959023222,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.054792027920485,
"inferenceMs": 46.54541704803705,
"masksPerSecond": 21.484392308010758,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.29462499031797,
"totalMs": 4562.690376071259,
"residentMemoryBytes": 2256306176,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4516.144959023222,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 436.5842909901403,
"inferenceMs": 483.16358297597617,
"masksPerSecond": 2.0696924090194146,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 432.6701249810867,
"totalMs": 4999.3085419991985,
"residentMemoryBytes": 2255634432,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4516.144959023222,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.20245899539441,
"inferenceMs": 61.98437500279397,
"masksPerSecond": 16.133098058259435,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.326791998930275,
"totalMs": 4578.129334026016,
"residentMemoryBytes": 2254946304,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4516.144959023222,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.04249997762963,
"inferenceMs": 62.024249986279756,
"masksPerSecond": 16.122726195338238,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.208083004225045,
"totalMs": 4578.169209009502,
"residentMemoryBytes": 2253766656,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4575.66712499829,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 511.72316598240286,
"inferenceMs": 592.1556670218706,
"masksPerSecond": 1.6887451318828062,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711893504,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 501.31187500664964,
"totalMs": 5167.822792020161,
"residentMemoryBytes": 509165568,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23666699075768668,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4575.66712499829,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 58.59766702633351,
"inferenceMs": 105.49895802978426,
"masksPerSecond": 9.47876660277234,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719446528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 51.34641600307077,
"totalMs": 4681.166083028074,
"residentMemoryBytes": 533463040,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37142609556820994,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4575.66712499829,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 360.16291700070724,
"inferenceMs": 390.81420801812783,
"masksPerSecond": 2.5587606066604804,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 352.5124169536866,
"totalMs": 4966.481333016418,
"residentMemoryBytes": 537542656,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.36998762069819263,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4575.66712499829,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.134333018679172,
"inferenceMs": 48.51012502331287,
"masksPerSecond": 20.614253200119,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.693666990846395,
"totalMs": 4624.177250021603,
"residentMemoryBytes": 537542656,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4575.66712499829,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 357.99458296969533,
"inferenceMs": 404.5281669823453,
"masksPerSecond": 2.4720157497553013,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 351.2765840278007,
"totalMs": 4980.195291980635,
"residentMemoryBytes": 536952832,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4575.66712499829,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.488499976228923,
"inferenceMs": 64.31491702096537,
"masksPerSecond": 15.54849242321218,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.968042002990842,
"totalMs": 4639.9820420192555,
"residentMemoryBytes": 539082752,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4575.66712499829,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.628082994837314,
"inferenceMs": 67.27312499424443,
"masksPerSecond": 14.864776983164605,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.739916983060539,
"totalMs": 4642.9402499925345,
"residentMemoryBytes": 541229056,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4455.478999996558,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 483.76724997069687,
"inferenceMs": 566.0961660323665,
"masksPerSecond": 1.7664843183955166,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711893504,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 474.00654200464487,
"totalMs": 5021.575166028924,
"residentMemoryBytes": 511197184,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23666699075768668,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4455.478999996558,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 55.111915979068726,
"inferenceMs": 105.8038339833729,
"masksPerSecond": 9.451453339178146,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719446528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 47.741582966409624,
"totalMs": 4561.282833979931,
"residentMemoryBytes": 535592960,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37142609556820994,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4455.478999996558,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 364.87770796520635,
"inferenceMs": 396.9544169958681,
"masksPerSecond": 2.519180936612198,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 356.9913330138661,
"totalMs": 4852.433416992426,
"residentMemoryBytes": 539738112,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.36998762069819263,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4455.478999996558,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 20.29441698687151,
"inferenceMs": 52.18229198362678,
"masksPerSecond": 19.16358906415551,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.323583010584116,
"totalMs": 4507.661291980185,
"residentMemoryBytes": 539738112,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4455.478999996558,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 358.9822080102749,
"inferenceMs": 407.7745830290951,
"masksPerSecond": 2.4523352891973875,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 352.04850003356114,
"totalMs": 4863.253583025653,
"residentMemoryBytes": 539131904,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4455.478999996558,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.038959009572864,
"inferenceMs": 65.73224999010563,
"masksPerSecond": 15.213232471892038,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.11125003080815,
"totalMs": 4521.2112499866635,
"residentMemoryBytes": 541261824,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4455.478999996558,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.872665998060256,
"inferenceMs": 68.81233298918232,
"masksPerSecond": 14.532278685525828,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.877582990564406,
"totalMs": 4524.29133298574,
"residentMemoryBytes": 543440896,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4684.5709580229595,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 464.52454203972593,
"inferenceMs": 543.4454999631271,
"masksPerSecond": 1.8401109220112233,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711893504,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 453.988625027705,
"totalMs": 5228.016457986087,
"residentMemoryBytes": 511459328,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23666699075768668,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4684.5709580229595,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 50.34112499561161,
"inferenceMs": 96.21720900759101,
"masksPerSecond": 10.393151186926504,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719446528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 43.24791696853936,
"totalMs": 4780.7881670305505,
"residentMemoryBytes": 536477696,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37142609556820994,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4684.5709580229595,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 356.5222910256125,
"inferenceMs": 386.8271670071408,
"masksPerSecond": 2.5851338408750903,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 349.07341696089134,
"totalMs": 5071.3981250301,
"residentMemoryBytes": 540295168,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.36998762069819263,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4684.5709580229595,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.456999969203025,
"inferenceMs": 48.29070798587054,
"masksPerSecond": 20.707917562372284,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.103458993602544,
"totalMs": 4732.86166600883,
"residentMemoryBytes": 540295168,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4684.5709580229595,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 355.7407920015976,
"inferenceMs": 401.48091700393707,
"masksPerSecond": 2.4907784097499053,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 349.10691605182365,
"totalMs": 5086.051875026897,
"residentMemoryBytes": 539705344,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4684.5709580229595,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.104500020854175,
"inferenceMs": 63.98012500721961,
"masksPerSecond": 15.6298538004913,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.309500026982278,
"totalMs": 4748.551083030179,
"residentMemoryBytes": 541851648,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4684.5709580229595,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.520249985158443,
"inferenceMs": 65.67587499739602,
"masksPerSecond": 15.226291237682771,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.828707996755838,
"totalMs": 4750.246833020356,
"residentMemoryBytes": 543997952,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4792.196042020805,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 491.0732079879381,
"inferenceMs": 573.487333022058,
"masksPerSecond": 1.7437176767800329,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711893504,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 481.27354198368266,
"totalMs": 5365.683375042863,
"residentMemoryBytes": 511967232,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23184137079062958,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4792.196042020805,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 53.468375001102686,
"inferenceMs": 99.9443749897182,
"masksPerSecond": 10.005565596892023,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719446528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 46.60283401608467,
"totalMs": 4892.140417010523,
"residentMemoryBytes": 536412160,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37162911611785093,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4792.196042020805,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 362.3618329875171,
"inferenceMs": 393.1593330344185,
"masksPerSecond": 2.543498057853447,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 354.8071249970235,
"totalMs": 5185.355375055224,
"residentMemoryBytes": 540655616,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3704481307254271,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4792.196042020805,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 16.902957984711975,
"inferenceMs": 47.14695899747312,
"masksPerSecond": 21.210275726449204,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 9.348667052108794,
"totalMs": 4839.343001018278,
"residentMemoryBytes": 540868608,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4792.196042020805,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 355.96641700249165,
"inferenceMs": 402.10904099512845,
"masksPerSecond": 2.4868876300946314,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 349.456875002943,
"totalMs": 5194.305083015934,
"residentMemoryBytes": 540295168,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4792.196042020805,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 17.2958750044927,
"inferenceMs": 62.97366600483656,
"masksPerSecond": 15.87965356698778,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.821167030371726,
"totalMs": 4855.169708025642,
"residentMemoryBytes": 542425088,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4792.196042020805,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 18.52662500459701,
"inferenceMs": 65.35829103086144,
"masksPerSecond": 15.300277657624362,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.651707987766713,
"totalMs": 4857.554333051667,
"residentMemoryBytes": 544587776,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4460.386000049766,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 468.28741702483967,
"inferenceMs": 548.4770420007408,
"masksPerSecond": 1.8232303695924785,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711893504,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 457.71662500919774,
"totalMs": 5008.863042050507,
"residentMemoryBytes": 513638400,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23184137079062958,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4460.386000049766,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 50.80691701732576,
"inferenceMs": 98.8403339870274,
"masksPerSecond": 10.117327205017821,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719446528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 43.83108299225569,
"totalMs": 4559.226334036794,
"residentMemoryBytes": 538083328,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37162911611785093,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4460.386000049766,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 358.3217079867609,
"inferenceMs": 389.32904094690457,
"masksPerSecond": 2.56852146854459,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 350.74579098727554,
"totalMs": 4849.715040996671,
"residentMemoryBytes": 541736960,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3704481307254271,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4460.386000049766,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 19.49391601374373,
"inferenceMs": 49.02737500378862,
"masksPerSecond": 20.396768130513298,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.966084013693035,
"totalMs": 4509.413375053555,
"residentMemoryBytes": 541736960,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4460.386000049766,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 357.90354100754485,
"inferenceMs": 404.4245000113733,
"masksPerSecond": 2.4726494066800546,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 351.2602500268258,
"totalMs": 4864.8105000611395,
"residentMemoryBytes": 541114368,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4460.386000049766,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 16.993124969303608,
"inferenceMs": 63.854208041448146,
"masksPerSecond": 15.660675007524862,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.366833012085408,
"totalMs": 4524.240208091214,
"residentMemoryBytes": 543244288,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4460.386000049766,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 18.85737502016127,
"inferenceMs": 67.24929198389873,
"masksPerSecond": 14.87004502946182,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.823749984614551,
"totalMs": 4527.635292033665,
"residentMemoryBytes": 545390592,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4535.761374980211,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 463.63612497225404,
"inferenceMs": 544.8559169890359,
"masksPerSecond": 1.8353476007495078,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711893504,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 453.7309579900466,
"totalMs": 5080.617291969247,
"residentMemoryBytes": 514654208,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23184137079062958,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4535.761374980211,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 50.00412499066442,
"inferenceMs": 97.08670899271965,
"masksPerSecond": 10.300071043452386,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719446528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 43.129292025696486,
"totalMs": 4632.848083972931,
"residentMemoryBytes": 539361280,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37162911611785093,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4535.761374980211,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 357.5546249630861,
"inferenceMs": 388.5344169684686,
"masksPerSecond": 2.573774564947112,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 350.0749579980038,
"totalMs": 4924.29579194868,
"residentMemoryBytes": 543227904,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3704481307254271,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4535.761374980211,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 19.403124984819442,
"inferenceMs": 48.94437501206994,
"masksPerSecond": 20.431357020155936,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.479249980766326,
"totalMs": 4584.705749992281,
"residentMemoryBytes": 543244288,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4535.761374980211,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 356.41429194947705,
"inferenceMs": 401.53695800108835,
"masksPerSecond": 2.49043078121165,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 349.83054199256003,
"totalMs": 4937.2983329813,
"residentMemoryBytes": 542654464,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4535.761374980211,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 19.055292010307312,
"inferenceMs": 63.95770801464096,
"masksPerSecond": 15.635332019263164,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.979791044723243,
"totalMs": 4599.719082994852,
"residentMemoryBytes": 544784384,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4535.761374980211,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 17.581416992470622,
"inferenceMs": 63.46474995370954,
"masksPerSecond": 15.756778380587468,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.996999975759536,
"totalMs": 4599.226124933921,
"residentMemoryBytes": 546930688,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4479.100624972489,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 529.4213339802809,
"inferenceMs": 609.037125017494,
"masksPerSecond": 1.6419360313565057,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4559142912,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 519.6524580242112,
"totalMs": 5088.137749989983,
"residentMemoryBytes": 492765184,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4479.100624972489,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 53.619417012669146,
"inferenceMs": 101.47187497932464,
"masksPerSecond": 9.854947493615887,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568793088,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 47.106417012400925,
"totalMs": 4580.572499951813,
"residentMemoryBytes": 515096576,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4479.100624972489,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 375.03179197665304,
"inferenceMs": 405.71966697461903,
"masksPerSecond": 2.464756040684017,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 367.57079197559506,
"totalMs": 4884.820291947108,
"residentMemoryBytes": 524156928,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4479.100624972489,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.320208007935435,
"inferenceMs": 48.85437502525747,
"masksPerSecond": 20.46899585723909,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.949041985441,
"totalMs": 4527.954999997746,
"residentMemoryBytes": 524189696,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4479.100624972489,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 375.9540000464767,
"inferenceMs": 422.1782079548575,
"masksPerSecond": 2.368667972807653,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 369.49358304264024,
"totalMs": 4901.278832927346,
"residentMemoryBytes": 524632064,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4479.100624972489,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.034875032957643,
"inferenceMs": 66.36179203633219,
"masksPerSecond": 15.068911934332837,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.632916972506791,
"totalMs": 4545.462417008821,
"residentMemoryBytes": 526794752,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4479.100624972489,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.482542025391012,
"inferenceMs": 64.03483304893598,
"masksPerSecond": 15.616500463674065,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.708042012993246,
"totalMs": 4543.135458021425,
"residentMemoryBytes": 528924672,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4387.93133298168,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 497.4010420264676,
"inferenceMs": 577.1342500229366,
"masksPerSecond": 1.7326991076344156,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4559142912,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 486.8512090179138,
"totalMs": 4965.065583004616,
"residentMemoryBytes": 492912640,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4387.93133298168,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 44.31841697078198,
"inferenceMs": 91.67558304034173,
"masksPerSecond": 10.908029890139353,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568793088,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 37.138417013920844,
"totalMs": 4479.606916022021,
"residentMemoryBytes": 515342336,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4387.93133298168,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 375.0511250109412,
"inferenceMs": 404.9683330231346,
"masksPerSecond": 2.469328879458022,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 367.70866601727903,
"totalMs": 4792.899666004814,
"residentMemoryBytes": 524288000,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4387.93133298168,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.881750002037734,
"inferenceMs": 49.85120799392462,
"masksPerSecond": 20.059694443550303,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.37295801891014,
"totalMs": 4437.782540975604,
"residentMemoryBytes": 524320768,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4387.93133298168,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 377.18337500700727,
"inferenceMs": 424.5984590379521,
"masksPerSecond": 2.355166342962673,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 370.6137500121258,
"totalMs": 4812.529792019632,
"residentMemoryBytes": 523763712,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4387.93133298168,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.269249994773418,
"inferenceMs": 64.63945802533999,
"masksPerSecond": 15.4704267416348,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.774334008805454,
"totalMs": 4452.57079100702,
"residentMemoryBytes": 525926400,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4387.93133298168,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.218624976929277,
"inferenceMs": 65.66275004297495,
"masksPerSecond": 15.229334734617726,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.381542008370161,
"totalMs": 4453.5940830246545,
"residentMemoryBytes": 528089088,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4474.526166974101,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 500.41266600601375,
"inferenceMs": 580.5262080393732,
"masksPerSecond": 1.7225751157339253,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4559142912,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 490.16595899593085,
"totalMs": 5055.052375013474,
"residentMemoryBytes": 492027904,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4474.526166974101,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 49.121999996714294,
"inferenceMs": 98.19420898566023,
"masksPerSecond": 10.18389995021025,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568793088,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 42.4662500154227,
"totalMs": 4572.720375959761,
"residentMemoryBytes": 514277376,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4474.526166974101,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 377.4574580020271,
"inferenceMs": 408.4703749977052,
"masksPerSecond": 2.448157960061652,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 370.025040989276,
"totalMs": 4882.996541971806,
"residentMemoryBytes": 523223040,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4474.526166974101,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.949458015616983,
"inferenceMs": 50.0263330177404,
"masksPerSecond": 19.989472337406358,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.634291022550315,
"totalMs": 4524.552499991842,
"residentMemoryBytes": 523255808,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4474.526166974101,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 375.1884579542093,
"inferenceMs": 421.54387501068413,
"masksPerSecond": 2.372232309091562,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 368.5335419722833,
"totalMs": 4896.070041984785,
"residentMemoryBytes": 522698752,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4474.526166974101,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.515125022735447,
"inferenceMs": 65.76062500244007,
"masksPerSecond": 15.206668123408114,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.746249965857714,
"totalMs": 4540.286791976541,
"residentMemoryBytes": 524861440,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4474.526166974101,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.89204198960215,
"inferenceMs": 66.35649997042492,
"masksPerSecond": 15.070113710724643,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.103167013265193,
"totalMs": 4540.882666944526,
"residentMemoryBytes": 527007744,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 7760.743332968559,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1920.4569170251489,
"inferenceMs": 3049.058042000979,
"masksPerSecond": 0.32797014232754274,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4559126528,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 1909.56170798745,
"totalMs": 10809.801374969538,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "acceptable"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 7760.743332968559,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 466.5965409949422,
"inferenceMs": 513.0449169664644,
"masksPerSecond": 1.9491470764641956,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568776704,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 459.8453330108896,
"totalMs": 8273.788249935023,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4730.376833002083,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 754.030124982819,
"inferenceMs": 831.3734159455635,
"masksPerSecond": 1.2028289344116794,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 744.5200830115937,
"totalMs": 5561.7502489476465,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4730.376833002083,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 25.273541978094727,
"inferenceMs": 70.95995801500976,
"masksPerSecond": 14.092454786803504,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 19.0122919739224,
"totalMs": 4801.336791017093,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2549199876464129,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4729.050041991286,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 499.3469579494558,
"inferenceMs": 578.0063339625485,
"masksPerSecond": 1.7300848472445878,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 491.8595839990303,
"totalMs": 5307.0563759538345,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23189856332357248,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4729.050041991286,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.903707990422845,
"inferenceMs": 63.93058400135487,
"masksPerSecond": 15.641965666680084,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.827790982555598,
"totalMs": 4792.980625992641,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25497860999267935,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4490.158875007182,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 2036.0011659795418,
"inferenceMs": 2113.9211659901775,
"masksPerSecond": 0.47305453774175726,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711877120,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 2018.7592080328614,
"totalMs": 6604.08004099736,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "acceptable"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23184137079062958,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4490.158875007182,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 498.7416659714654,
"inferenceMs": 546.6157500050031,
"masksPerSecond": 1.8294386870316985,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 488.5219170246273,
"totalMs": 5036.774625012185,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25436379026354317,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4617.2077499795705,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 2155.9302500099875,
"inferenceMs": 2237.676332995761,
"masksPerSecond": 0.4468921556055508,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711877120,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 2136.923125013709,
"totalMs": 6854.884082975332,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "acceptable"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.23666699075768668,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4617.2077499795705,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 472.67958300653845,
"inferenceMs": 517.9352920386009,
"masksPerSecond": 1.9307431167009017,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719430144,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 465.7067919615656,
"totalMs": 5135.143042018171,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 6307.8163750469685,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 607.1678329608403,
"inferenceMs": 667.8932920331135,
"masksPerSecond": 1.4972451616573819,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4559142912,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 586.4486659993418,
"totalMs": 6975.709667080082,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 6307.8163750469685,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 58.154416969046,
"inferenceMs": 87.9744160338305,
"masksPerSecond": 11.366941039034016,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4568793088,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 50.472833041567355,
"totalMs": 6395.790791080799,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4440.892792015802,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 756.6732080304064,
"inferenceMs": 807.7346669742838,
"masksPerSecond": 1.2380303098119194,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 745.5031669815071,
"totalMs": 5248.627458990086,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4440.892792015802,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.293291975278407,
"inferenceMs": 52.10604198509827,
"masksPerSecond": 19.19163233096823,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.098166001029313,
"totalMs": 4492.9988340009,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3716687298836346,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4499.261167016812,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 488.1767500191927,
"inferenceMs": 538.6726669967175,
"masksPerSecond": 1.856414964537441,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 479.97262503486127,
"totalMs": 5037.933834013529,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3706759098786828,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4499.261167016812,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.155875033698976,
"inferenceMs": 48.30229200888425,
"masksPerSecond": 20.70295131783953,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 13.232707977294922,
"totalMs": 4547.563459025696,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37162911611785093,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4754.259291978087,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 496.7593750334345,
"inferenceMs": 547.9165409924462,
"masksPerSecond": 1.8250954756516218,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711893504,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 485.74787500547245,
"totalMs": 5302.1758329705335,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3704481307254271,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4754.259291978087,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 52.70783300511539,
"inferenceMs": 82.58329197997227,
"masksPerSecond": 12.108986890986563,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719446528,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 45.35358300199732,
"totalMs": 4836.84258395806,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37142609556820994,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4716.300999978557,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 490.482792025432,
"inferenceMs": 541.6860420373268,
"masksPerSecond": 1.8460878117496176,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3711893504,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 477.7204589918256,
"totalMs": 5257.987042015884,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.36998762069819263,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4716.300999978557,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 54.57025003852323,
"inferenceMs": 84.40966700436547,
"masksPerSecond": 11.846984302737296,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3719446528,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 46.89612495712936,
"totalMs": 4800.710666982923,
"artifactRevision": "79c09c1ce6b4ae51f00634ed171d9b8e888f6911",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
}
]SlimSAM 50 Uniform Desktop
125 configurations testedQuality criteria
{ "metric": "Independent foreground/background probe accuracy and whole-subject area bounds", "minimumProbeAccuracy": 0.9, "requirePromptSemantics": true, "frozenPolicy": "research/slimsam-family-expansion/quality-gates.json" }
| Precision | Device | Result | Platform | Inference / load | |
|---|---|---|---|---|---|
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 47 ms / 4829 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 48 ms / 4399 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 48 ms / 4534 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 49 ms / 4405 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 49 ms / 4779 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 50 ms / 4484 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 50 ms / 4708 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 50 ms / 4549 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 51 ms / 4496 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 64 ms / 4829 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 65 ms / 4405 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 65 ms / 4534 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 65 ms / 4496 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 65 ms / 4484 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 65 ms / 4405 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 65 ms / 4399 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 65 ms / 4484 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4829 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4549 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4399 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4779 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4549 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4496 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 66 ms / 4534 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 67 ms / 4779 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 67 ms / 4708 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 69 ms / 4708 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 77 ms / 4566 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 81 ms / 4678 ms |
BF16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 88 ms / 4581 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 91 ms / 4399 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 93 ms / 4484 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 95 ms / 4562 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 96 ms / 4829 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 96 ms / 5156 ms |
FP32 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 98 ms / 4534 ms |
FP16 | mpsAccelerated | Passed | Very fast | macos · aarch64 | 98 ms / 4708 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 101 ms / 4549 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 102 ms / 4779 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 104 ms / 4496 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 105 ms / 4677 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 107 ms / 4405 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 409 ms / 4549 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 411 ms / 4496 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 413 ms / 4829 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 414 ms / 4405 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 416 ms / 4708 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 417 ms / 4484 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 425 ms / 4829 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 427 ms / 4405 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 427 ms / 4549 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 430 ms / 4484 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 430 ms / 4708 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 431 ms / 4496 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 433 ms / 4534 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 434 ms / 4779 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 436 ms / 4399 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 455 ms / 4399 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 455 ms / 4534 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 458 ms / 4779 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 555 ms / 4678 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 558 ms / 4581 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 568 ms / 4708 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 569 ms / 4549 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 572 ms / 4829 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 582 ms / 4566 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 585 ms / 4496 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 595 ms / 4484 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 602 ms / 4534 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 609 ms / 4399 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 627 ms / 4779 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 667 ms / 4405 ms |
FP32 | mpsAccelerated | Passed | Fast | macos · aarch64 | 855 ms / 4562 ms |
FP16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 990 ms / 5156 ms |
BF16 | mpsAccelerated | Passed | Fast | macos · aarch64 | 991 ms / 4677 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 47 ms / 4450 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 47 ms / 4404 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 47 ms / 4408 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 47 ms / 4338 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 52 ms / 4534 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 53 ms / 4420 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 54 ms / 4377 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 55 ms / 6010 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 62 ms / 4450 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 62 ms / 4338 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 62 ms / 4338 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 62 ms / 4450 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 63 ms / 4450 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 63 ms / 4404 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 63 ms / 4404 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 63 ms / 4722 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 64 ms / 4338 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 66 ms / 4404 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 67 ms / 4377 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 68 ms / 6010 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 69 ms / 4420 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 69 ms / 6010 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 69 ms / 6010 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 69 ms / 4420 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 70 ms / 4377 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 71 ms / 4377 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 71 ms / 4420 ms |
FP32 | cpu | Passed | Very fast | macos · aarch64 | 71 ms / 4508 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 580 ms / 4450 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 583 ms / 4338 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 590 ms / 4450 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 595 ms / 4404 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 601 ms / 4338 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 602 ms / 4404 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 638 ms / 4408 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 667 ms / 4450 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 672 ms / 4338 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 693 ms / 4404 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 717 ms / 4722 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 838 ms / 4420 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 846 ms / 6010 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 857 ms / 4377 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 858 ms / 4377 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 865 ms / 4420 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 874 ms / 6010 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 939 ms / 4534 ms |
FP32 | cpu | Passed | Fast | macos · aarch64 | 967 ms / 4420 ms |
FP32 | cpu | Passed | Moderate | macos · aarch64 | 1012 ms / 4377 ms |
FP32 | cpu | Passed | Moderate | macos · aarch64 | 1099 ms / 4508 ms |
FP32 | cpu | Passed | Moderate | macos · aarch64 | 1149 ms / 6010 ms |
Test setup & sample details
- Quality gate
- { "metric": "Independent foreground/background probe accuracy and whole-subject area bounds", "minimumProbeAccuracy": 0.9, "requirePromptSemantics": true, "frozenPolicy": "research/slimsam-family-expansion/quality-gates.json" }
- Tested
- 2026-11-13T10:43:24.851097+00:00
- Environment
- { "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max;Mac15,11;96GiB", "osVersion": "27.2", "runtime": "Transformers5.22.1/Torch2.10.0/Pillow12.0.0; MPS CPU-op fallback disabled", "transport": "Verified local original provider snapshots; no network-transfer timing", "concurrentUploadActivity": "none; uploads excluded" }
- Runtime
- transformers 5.22.1
- Artifact revision
- d67c795ed0caaec7423d799da25cdeded48aac40
- 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": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 6009.678458038252,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1060.492249962408,
"inferenceMs": 1148.8740830100141,
"masksPerSecond": 0.8704174067361945,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 1042.7828329848126,
"totalMs": 7158.552541048266,
"residentMemoryBytes": 2332246016,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "acceptable"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324556185944363,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 6009.678458038252,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.947208024561405,
"inferenceMs": 69.00545803364366,
"masksPerSecond": 14.491607308981983,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 17.044125008396804,
"totalMs": 6078.6839160718955,
"residentMemoryBytes": 2338127872,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 6009.678458038252,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 815.6332080252469,
"inferenceMs": 846.482458000537,
"masksPerSecond": 1.1813593897292147,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 807.65687499661,
"totalMs": 6856.160916038789,
"residentMemoryBytes": 2338340864,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 6009.678458038252,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.027791001368314,
"inferenceMs": 54.54462498892099,
"masksPerSecond": 18.333612160742113,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.01808296982199,
"totalMs": 6064.223083027173,
"residentMemoryBytes": 2342551552,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 6009.678458038252,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 827.713250007946,
"inferenceMs": 874.3083750014193,
"masksPerSecond": 1.143761204390129,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 821.891916973982,
"totalMs": 6883.986833039671,
"residentMemoryBytes": 2341928960,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 6009.678458038252,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.103417003061622,
"inferenceMs": 69.02920902939513,
"masksPerSecond": 14.486621157344622,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 18.329832993913442,
"totalMs": 6078.707667067647,
"residentMemoryBytes": 2340749312,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 6009.678458038252,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.90045895986259,
"inferenceMs": 67.89066700730473,
"masksPerSecond": 14.729565109330927,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.06879202881828,
"totalMs": 6077.569125045557,
"residentMemoryBytes": 2342879232,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4419.760124990717,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 889.4727909937501,
"inferenceMs": 966.9880000292324,
"masksPerSecond": 1.0341389965229866,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 880.1771670114249,
"totalMs": 5386.748125019949,
"residentMemoryBytes": 2330755072,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324556185944363,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4419.760124990717,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.951166971120983,
"inferenceMs": 69.12837497657165,
"masksPerSecond": 14.465839828274724,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.13825000822544,
"totalMs": 4488.888499967288,
"residentMemoryBytes": 2336718848,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4419.760124990717,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 808.0707910121419,
"inferenceMs": 838.2090419763699,
"masksPerSecond": 1.193019819545434,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 801.0591660276987,
"totalMs": 5257.9691669670865,
"residentMemoryBytes": 2336915456,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4419.760124990717,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.781916020903736,
"inferenceMs": 52.799833996687084,
"masksPerSecond": 18.9394534850762,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 15.736708010081202,
"totalMs": 4472.559958987404,
"residentMemoryBytes": 2341126144,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4419.760124990717,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 818.6026669573039,
"inferenceMs": 865.3069590218365,
"masksPerSecond": 1.1556592600739322,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 812.7426250139251,
"totalMs": 5285.067084012553,
"residentMemoryBytes": 2340536320,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4419.760124990717,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.040958971250802,
"inferenceMs": 68.51545802783221,
"masksPerSecond": 14.595246514936557,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 17.305249988567084,
"totalMs": 4488.275583018549,
"residentMemoryBytes": 2339356672,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4419.760124990717,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 22.246334003284574,
"inferenceMs": 70.69870794657618,
"masksPerSecond": 14.144530063486519,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.204125015065074,
"totalMs": 4490.458832937293,
"residentMemoryBytes": 2341486592,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4377.071207971312,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 933.4469999885187,
"inferenceMs": 1011.6507500060834,
"masksPerSecond": 0.9884834267102423,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 924.2101249983534,
"totalMs": 5388.7219579773955,
"residentMemoryBytes": 2330607616,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "acceptable"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324556185944363,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4377.071207971312,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.476916016079485,
"inferenceMs": 70.5118749756366,
"masksPerSecond": 14.182008354557611,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 17.348500026855618,
"totalMs": 4447.583082946949,
"residentMemoryBytes": 2336604160,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4377.071207971312,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 827.4463340057991,
"inferenceMs": 858.3184580202214,
"masksPerSecond": 1.1650687348686153,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 820.1485410099849,
"totalMs": 5235.389665991534,
"residentMemoryBytes": 2336817152,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4377.071207971312,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.111208982300013,
"inferenceMs": 53.670457971747965,
"masksPerSecond": 18.6322240910707,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.020916984416544,
"totalMs": 4430.74166594306,
"residentMemoryBytes": 2341879808,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4377.071207971312,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 809.261999966111,
"inferenceMs": 856.774999992922,
"masksPerSecond": 1.1671675760943787,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 803.4241659916006,
"totalMs": 5233.846207964234,
"residentMemoryBytes": 2341191680,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4377.071207971312,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.134165988769382,
"inferenceMs": 66.97970803361386,
"masksPerSecond": 14.929894879478253,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 15.361541998572648,
"totalMs": 4444.050916004926,
"residentMemoryBytes": 2340012032,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4377.071207971312,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 21.756375033874065,
"inferenceMs": 69.79770795442164,
"masksPerSecond": 14.327118028761152,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 15.818459040019661,
"totalMs": 4446.868915925734,
"residentMemoryBytes": 2342141952,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.1744580110535,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 594.8795829899609,
"inferenceMs": 672.2347500035539,
"masksPerSecond": 1.4875755827777621,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 587.7328329952434,
"totalMs": 5010.409208014607,
"residentMemoryBytes": 2331770880,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324556185944363,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.1744580110535,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.239500011783093,
"inferenceMs": 64.42279199836776,
"masksPerSecond": 15.52245671105556,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 13.176958018448204,
"totalMs": 4402.597250009421,
"residentMemoryBytes": 2339438592,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.1744580110535,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 552.9150410438888,
"inferenceMs": 583.4623330156319,
"masksPerSecond": 1.7139066969953112,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 548.1066249776632,
"totalMs": 4921.636791026685,
"residentMemoryBytes": 2340388864,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.1744580110535,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.143333971034735,
"inferenceMs": 47.448333993088454,
"masksPerSecond": 21.07555557473661,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.326041993219405,
"totalMs": 4385.622792004142,
"residentMemoryBytes": 2343452672,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.1744580110535,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 555.5538330227137,
"inferenceMs": 601.1989170219749,
"masksPerSecond": 1.6633429829738835,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 551.6330829705112,
"totalMs": 4939.373375033028,
"residentMemoryBytes": 2343354368,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.1744580110535,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 15.995708003174514,
"inferenceMs": 62.04454100225121,
"masksPerSecond": 16.11745342694559,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.005625001620501,
"totalMs": 4400.218999013305,
"residentMemoryBytes": 2342174720,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4338.1744580110535,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.177250014152378,
"inferenceMs": 62.13174998993054,
"masksPerSecond": 16.09483074534463,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.417917023412883,
"totalMs": 4400.306208000984,
"residentMemoryBytes": 2346401792,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4449.760083050933,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 589.1903750016354,
"inferenceMs": 667.4361249897629,
"masksPerSecond": 1.4982707146924927,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 582.1897499845363,
"totalMs": 5117.196208040696,
"residentMemoryBytes": 2334769152,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324556185944363,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4449.760083050933,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.87925000442192,
"inferenceMs": 62.023333040997386,
"masksPerSecond": 16.122964551727662,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.795207963790745,
"totalMs": 4511.78341609193,
"residentMemoryBytes": 2340323328,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4449.760083050933,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 549.9578750459477,
"inferenceMs": 580.3572500008158,
"masksPerSecond": 1.7230766049680508,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 545.1474169967696,
"totalMs": 5030.117333051749,
"residentMemoryBytes": 2342649856,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4449.760083050933,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.06424995791167,
"inferenceMs": 46.648250019643456,
"masksPerSecond": 21.437031390864664,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.408207985572517,
"totalMs": 4496.408333070576,
"residentMemoryBytes": 2344796160,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4449.760083050933,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 544.8006669757888,
"inferenceMs": 590.3891670168377,
"masksPerSecond": 1.6937980164047968,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 540.9938750090078,
"totalMs": 5040.1492500677705,
"residentMemoryBytes": 2344108032,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4449.760083050933,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.240250028204173,
"inferenceMs": 62.39337503211573,
"masksPerSecond": 16.02734263830527,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.331875041127205,
"totalMs": 4512.1534580830485,
"residentMemoryBytes": 2342928384,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4449.760083050933,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.470542002934963,
"inferenceMs": 62.58495897054672,
"masksPerSecond": 15.97828002844282,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.501582968980074,
"totalMs": 4512.3450420214795,
"residentMemoryBytes": 2345058304,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4403.571792005096,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 614.7286249906756,
"inferenceMs": 693.0762919946574,
"masksPerSecond": 1.4428426012409448,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 607.401417044457,
"totalMs": 5096.648083999753,
"residentMemoryBytes": 2332065792,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324556185944363,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4403.571792005096,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.6627499749884,
"inferenceMs": 65.69208396831527,
"masksPerSecond": 15.222534277985789,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 13.500000000931323,
"totalMs": 4469.263875973411,
"residentMemoryBytes": 2339684352,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4403.571792005096,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 565.5894579831511,
"inferenceMs": 595.3897089930251,
"masksPerSecond": 1.6795721943049486,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 560.6117090210319,
"totalMs": 4998.961500998121,
"residentMemoryBytes": 2342092800,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4403.571792005096,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.817625029943883,
"inferenceMs": 46.96104198228568,
"masksPerSecond": 21.294246417641524,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.135499971918762,
"totalMs": 4450.532833987381,
"residentMemoryBytes": 2342109184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4403.571792005096,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 551.4532499946654,
"inferenceMs": 601.9793330342509,
"masksPerSecond": 1.6611865974859021,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 547.3216249956749,
"totalMs": 5005.5511250393465,
"residentMemoryBytes": 2341421056,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4403.571792005096,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.475583019200712,
"inferenceMs": 62.65729101141915,
"masksPerSecond": 15.959834583620161,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.372291006613523,
"totalMs": 4466.229083016515,
"residentMemoryBytes": 2340241408,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4403.571792005096,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 16.229416010901332,
"inferenceMs": 62.689166981726885,
"masksPerSecond": 15.951719382257664,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 12.261042022146285,
"totalMs": 4466.2609589868225,
"residentMemoryBytes": 2342371328,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4405.011041963007,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 590.0517919799313,
"inferenceMs": 666.7003750335425,
"masksPerSecond": 1.4999241600091926,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753836544,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 579.3799590319395,
"totalMs": 5071.71141699655,
"residentMemoryBytes": 542834688,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24240197199853586,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4405.011041963007,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 59.49349998263642,
"inferenceMs": 106.74683400429785,
"masksPerSecond": 9.367959334135735,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761389568,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 52.1138749900274,
"totalMs": 4511.757875967305,
"residentMemoryBytes": 567476224,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3708541718247091,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4405.011041963007,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 383.64449999062344,
"inferenceMs": 413.7815000140108,
"masksPerSecond": 2.416734435846309,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 376.11345801269636,
"totalMs": 4818.792541977018,
"residentMemoryBytes": 571637760,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.36806635305768753,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4405.011041963007,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.228416971396655,
"inferenceMs": 49.17000001296401,
"masksPerSecond": 20.33760422485953,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.80250000813976,
"totalMs": 4454.181041975971,
"residentMemoryBytes": 571637760,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4405.011041963007,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 381.01070799166337,
"inferenceMs": 427.33520799083635,
"masksPerSecond": 2.3400833380933213,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 374.39387501217425,
"totalMs": 4832.346249953844,
"residentMemoryBytes": 570687488,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4405.011041963007,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 17.43733399780467,
"inferenceMs": 64.5214170217514,
"masksPerSecond": 15.498729664645785,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.022499995306134,
"totalMs": 4469.532458984759,
"residentMemoryBytes": 572817408,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4405.011041963007,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.405332986731082,
"inferenceMs": 65.27237500995398,
"masksPerSecond": 15.320416942810812,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.781417008023709,
"totalMs": 4470.283416972961,
"residentMemoryBytes": 574963712,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4549.3118750164285,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 489.4112499896437,
"inferenceMs": 569.1737500019372,
"masksPerSecond": 1.7569327468046383,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753836544,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 478.8086249609478,
"totalMs": 5118.485625018366,
"residentMemoryBytes": 542769152,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24240197199853586,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4549.3118750164285,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 51.493374980054796,
"inferenceMs": 100.62649997416884,
"masksPerSecond": 9.937740061084341,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761389568,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 44.60125003242865,
"totalMs": 4649.938374990597,
"residentMemoryBytes": 567377920,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3708541718247091,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4549.3118750164285,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 378.4853750257753,
"inferenceMs": 408.94329198636115,
"masksPerSecond": 2.445326820603164,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 371.41799996607006,
"totalMs": 4958.25516700279,
"residentMemoryBytes": 571228160,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.36806635305768753,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4549.3118750164285,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 20.352416031528264,
"inferenceMs": 50.36616598954424,
"masksPerSecond": 19.854598426403847,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.53450004151091,
"totalMs": 4599.678041005973,
"residentMemoryBytes": 571228160,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4549.3118750164285,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 380.9333340032026,
"inferenceMs": 427.39791702479124,
"masksPerSecond": 2.3397399944323896,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 374.37725003110245,
"totalMs": 4976.70979204122,
"residentMemoryBytes": 570277888,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4549.3118750164285,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.041374976746738,
"inferenceMs": 65.54941600188613,
"masksPerSecond": 15.255666045464476,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.440750036854297,
"totalMs": 4614.861291018315,
"residentMemoryBytes": 572407808,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4549.3118750164285,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.562000012025237,
"inferenceMs": 65.77895797090605,
"masksPerSecond": 15.20242993880047,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.852541996631771,
"totalMs": 4615.090832987335,
"residentMemoryBytes": 574570496,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4496.459375019185,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 504.30849997792393,
"inferenceMs": 585.3195409872569,
"masksPerSecond": 1.7084685030561302,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753836544,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 493.9337919931859,
"totalMs": 5081.778916006442,
"residentMemoryBytes": 545193984,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24240197199853586,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4496.459375019185,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 53.615375014487654,
"inferenceMs": 103.89450000366196,
"masksPerSecond": 9.625148587892074,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761389568,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 46.69899999862537,
"totalMs": 4600.353875022847,
"residentMemoryBytes": 569917440,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3708541718247091,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4496.459375019185,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 380.5019169813022,
"inferenceMs": 410.98508395953104,
"masksPerSecond": 2.4331783294073714,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 372.5320830126293,
"totalMs": 4907.444458978716,
"residentMemoryBytes": 574029824,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.36806635305768753,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4496.459375019185,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 19.885291054379195,
"inferenceMs": 50.66675000125542,
"masksPerSecond": 19.736809642916153,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.223790981806815,
"totalMs": 4547.126125020441,
"residentMemoryBytes": 574046208,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4496.459375019185,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 384.2127920361236,
"inferenceMs": 431.0950840008445,
"masksPerSecond": 2.319673865726664,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 377.3635830148123,
"totalMs": 4927.55445902003,
"residentMemoryBytes": 573128704,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4496.459375019185,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.421709013637155,
"inferenceMs": 64.98754100175574,
"masksPerSecond": 15.3875648252791,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.773582955356687,
"totalMs": 4561.446916020941,
"residentMemoryBytes": 575258624,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4496.459375019185,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 18.561833014246076,
"inferenceMs": 65.87445898912847,
"masksPerSecond": 15.180390326469839,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.753916973248124,
"totalMs": 4562.333834008314,
"residentMemoryBytes": 577404928,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4483.865291986149,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 516.9243339914829,
"inferenceMs": 594.9657920282334,
"masksPerSecond": 1.68076890032788,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753836544,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 506.1421660357155,
"totalMs": 5078.831084014382,
"residentMemoryBytes": 542883840,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2437045319363104,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4483.865291986149,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 47.84616600954905,
"inferenceMs": 93.16950000356883,
"masksPerSecond": 10.733126183586853,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761389568,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 40.914665965829045,
"totalMs": 4577.034791989718,
"residentMemoryBytes": 567541760,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37081703391928694,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4483.865291986149,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 385.6438329676166,
"inferenceMs": 416.59416700713336,
"masksPerSecond": 2.4004176707132747,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 377.96275003347546,
"totalMs": 4900.459458993282,
"residentMemoryBytes": 571506688,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3678583807873236,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4483.865291986149,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 19.50116700027138,
"inferenceMs": 49.63416699320078,
"masksPerSecond": 20.147411764500585,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.9180420297198,
"totalMs": 4533.49945897935,
"residentMemoryBytes": 571506688,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4483.865291986149,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 383.51529196370393,
"inferenceMs": 430.1311250310391,
"masksPerSecond": 2.324872444252524,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 376.94554100744426,
"totalMs": 4913.996417017188,
"residentMemoryBytes": 570556416,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4483.865291986149,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 18.318874994292855,
"inferenceMs": 65.04129100358114,
"masksPerSecond": 15.374848570348036,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.962249991483986,
"totalMs": 4548.90658298973,
"residentMemoryBytes": 572686336,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4483.865291986149,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 18.371416023001075,
"inferenceMs": 65.4222079901956,
"masksPerSecond": 15.285329412144932,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.68420899193734,
"totalMs": 4549.287499976344,
"residentMemoryBytes": 574832640,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4829.261125007179,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 489.6799169946462,
"inferenceMs": 571.9362079980783,
"masksPerSecond": 1.7484467428636026,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753836544,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 479.57412496907637,
"totalMs": 5401.197333005257,
"residentMemoryBytes": 544194560,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2437045319363104,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4829.261125007179,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 49.80529099702835,
"inferenceMs": 95.67720897030085,
"masksPerSecond": 10.45180990083448,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761389568,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 43.10395801439881,
"totalMs": 4924.9383339774795,
"residentMemoryBytes": 569262080,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37081703391928694,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4829.261125007179,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 382.08787504117936,
"inferenceMs": 412.6394579652697,
"masksPerSecond": 2.423423113560231,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 374.62012498872355,
"totalMs": 5241.900582972448,
"residentMemoryBytes": 573161472,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3678583807873236,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4829.261125007179,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 17.534874961711466,
"inferenceMs": 47.483625006861985,
"masksPerSecond": 21.059891696463513,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.131124989129603,
"totalMs": 4876.744750014041,
"residentMemoryBytes": 573177856,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4829.261125007179,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 378.8422499783337,
"inferenceMs": 425.3620419767685,
"masksPerSecond": 2.350938497832902,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 372.3006669897586,
"totalMs": 5254.623166983947,
"residentMemoryBytes": 572243968,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4829.261125007179,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 17.923457955475897,
"inferenceMs": 63.747999956831336,
"masksPerSecond": 15.686766654282122,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.481125024147332,
"totalMs": 4893.00912496401,
"residentMemoryBytes": 574373888,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4829.261125007179,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 18.504542007576674,
"inferenceMs": 65.51024998771027,
"masksPerSecond": 15.264786811034917,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.806249967776239,
"totalMs": 4894.771374994889,
"residentMemoryBytes": 576520192,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4708.256541984156,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 487.885374983307,
"inferenceMs": 568.1003330391832,
"masksPerSecond": 1.760252444581876,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753836544,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 477.79766598250717,
"totalMs": 5276.356875023339,
"residentMemoryBytes": 542523392,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2437045319363104,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4708.256541984156,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 50.07037502946332,
"inferenceMs": 97.93408296536654,
"masksPerSecond": 10.210949750289085,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761389568,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 43.34700002800673,
"totalMs": 4806.190624949522,
"residentMemoryBytes": 567017472,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37081703391928694,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4708.256541984156,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 385.84595802240074,
"inferenceMs": 416.142791043967,
"masksPerSecond": 2.403021322299793,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 378.2964999554679,
"totalMs": 5124.399333028123,
"residentMemoryBytes": 570720256,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3678583807873236,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4708.256541984156,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 19.59224999882281,
"inferenceMs": 49.988333019427955,
"masksPerSecond": 20.004667881430457,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.050458986777812,
"totalMs": 4758.244875003584,
"residentMemoryBytes": 570720256,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4708.256541984156,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 382.0034160162322,
"inferenceMs": 430.278540996369,
"masksPerSecond": 2.3240759292442585,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 375.3359580296092,
"totalMs": 5138.535082980525,
"residentMemoryBytes": 569786368,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4708.256541984156,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 18.476415949407965,
"inferenceMs": 67.13958299951628,
"masksPerSecond": 14.894343326606672,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.758040986023843,
"totalMs": 4775.396124983672,
"residentMemoryBytes": 571916288,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4708.256541984156,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 19.03591601876542,
"inferenceMs": 68.66345799062401,
"masksPerSecond": 14.563787337022116,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.801333981566131,
"totalMs": 4776.91999997478,
"residentMemoryBytes": 574078976,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4778.857665951364,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 545.200707972981,
"inferenceMs": 626.7353749717586,
"masksPerSecond": 1.5955697411288794,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4584308736,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 513.9639170374721,
"totalMs": 5405.593040923122,
"residentMemoryBytes": 491192320,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324413204612005,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4778.857665951364,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 54.11075003212318,
"inferenceMs": 101.55179200228304,
"masksPerSecond": 9.847192061145691,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593958912,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 47.682665986940265,
"totalMs": 4880.409457953647,
"residentMemoryBytes": 514211840,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4778.857665951364,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 403.292374976445,
"inferenceMs": 433.78825002582744,
"masksPerSecond": 2.305272215050686,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 395.7151250215247,
"totalMs": 5212.645915977191,
"residentMemoryBytes": 521601024,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4778.857665951364,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.816165964584798,
"inferenceMs": 49.41824998240918,
"masksPerSecond": 20.235439343885265,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.688875023741275,
"totalMs": 4828.275915933773,
"residentMemoryBytes": 522108928,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4778.857665951364,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 405.21670802263543,
"inferenceMs": 457.6109159970656,
"masksPerSecond": 2.1852625561196457,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 398.3664999832399,
"totalMs": 5236.468581948429,
"residentMemoryBytes": 521535488,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4778.857665951364,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 20.545875013340265,
"inferenceMs": 66.70595897594467,
"masksPerSecond": 14.991164437956996,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 13.414208951871842,
"totalMs": 4845.563624927308,
"residentMemoryBytes": 524353536,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4778.857665951364,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.9094579545781,
"inferenceMs": 65.7071250025183,
"masksPerSecond": 15.219049683906789,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.0630839955993,
"totalMs": 4844.564790953882,
"residentMemoryBytes": 526483456,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4534.142374992371,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 522.0927919726819,
"inferenceMs": 602.2129170014523,
"masksPerSecond": 1.6605422629909952,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4584308736,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 490.498083003331,
"totalMs": 5136.355291993823,
"residentMemoryBytes": 493076480,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324413204612005,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4534.142374992371,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 49.874667020048946,
"inferenceMs": 97.88783302064985,
"masksPerSecond": 10.215774209538848,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593958912,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 43.105625023599714,
"totalMs": 4632.0302080130205,
"residentMemoryBytes": 515375104,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4534.142374992371,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 401.3789999880828,
"inferenceMs": 432.5006669969298,
"masksPerSecond": 2.3121351625732838,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 393.9537910046056,
"totalMs": 4966.6430419893,
"residentMemoryBytes": 524337152,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4534.142374992371,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.933832994662225,
"inferenceMs": 48.38187497807667,
"masksPerSecond": 20.668897194520284,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.380374955479056,
"totalMs": 4582.524249970447,
"residentMemoryBytes": 524369920,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4534.142374992371,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 403.0637079849839,
"inferenceMs": 455.1123749697581,
"masksPerSecond": 2.1972595231374434,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 396.00529201561585,
"totalMs": 4989.254749962129,
"residentMemoryBytes": 523829248,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4534.142374992371,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.24150001630187,
"inferenceMs": 64.83220803784207,
"masksPerSecond": 15.424432242324796,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 11.800166044849902,
"totalMs": 4598.974583030213,
"residentMemoryBytes": 525991936,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4534.142374992371,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.127750012557954,
"inferenceMs": 66.09241699334234,
"masksPerSecond": 15.13032879552177,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.470666028093547,
"totalMs": 4600.234791985713,
"residentMemoryBytes": 528121856,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4399.3548329453915,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 528.0872089788318,
"inferenceMs": 608.62149996683,
"masksPerSecond": 1.6430573025344983,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4584308736,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 496.48566596442834,
"totalMs": 5007.976332912222,
"residentMemoryBytes": 492290048,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324413204612005,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4399.3548329453915,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 43.71233400888741,
"inferenceMs": 90.77629097737372,
"masksPerSecond": 11.016092299356593,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593958912,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 36.95866698399186,
"totalMs": 4490.131123922765,
"residentMemoryBytes": 514703360,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4399.3548329453915,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 405.80191696062684,
"inferenceMs": 436.057000013534,
"masksPerSecond": 2.293278172277851,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 398.1054580071941,
"totalMs": 4835.4118329589255,
"residentMemoryBytes": 523747328,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4399.3548329453915,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.894916993100196,
"inferenceMs": 47.872458992060274,
"masksPerSecond": 20.8888371530247,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.440833983011544,
"totalMs": 4447.227291937452,
"residentMemoryBytes": 523780096,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4399.3548329453915,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 403.5727910231799,
"inferenceMs": 454.65208298992366,
"masksPerSecond": 2.1994840393641457,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 396.5964159579016,
"totalMs": 4854.006915935315,
"residentMemoryBytes": 523223040,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4399.3548329453915,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 19.01804201770574,
"inferenceMs": 65.28275000164285,
"masksPerSecond": 15.317982161824292,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 12.60383299086243,
"totalMs": 4464.637582947034,
"residentMemoryBytes": 525385728,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4399.3548329453915,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.108541971538216,
"inferenceMs": 65.56683301459998,
"masksPerSecond": 15.251613567751956,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference RSS/Metal driver allocation, not peak or validated minimum",
"modelInferenceMs": 10.653416975401342,
"totalMs": 4464.9216659599915,
"residentMemoryBytes": 527515648,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4561.913874989841,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 774.1430419846438,
"inferenceMs": 854.722749965731,
"masksPerSecond": 1.1699700283397085,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4584308736,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 743.4937080251984,
"totalMs": 5416.636624955572,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324413204612005,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4561.913874989841,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 49.37591700581834,
"inferenceMs": 95.17487499397248,
"masksPerSecond": 10.50697466178265,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593942528,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 43.148791999556124,
"totalMs": 4657.088749983814,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4508.377457968891,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 1012.6754580414854,
"inferenceMs": 1099.1950419847853,
"masksPerSecond": 0.9097566508254334,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 1001.5830830088817,
"totalMs": 5607.572499953676,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "acceptable"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324556185944363,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4508.377457968891,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 24.72741698147729,
"inferenceMs": 71.25687500229105,
"masksPerSecond": 14.033733586658803,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 18.383791961241513,
"totalMs": 4579.634332971182,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2548056025805271,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4721.630625019316,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 635.3316250024363,
"inferenceMs": 717.0942919910885,
"masksPerSecond": 1.3945167478929357,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 627.0873749745078,
"totalMs": 5438.724917010404,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24324556185944363,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4721.630625019316,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 17.90620799874887,
"inferenceMs": 63.22349997935817,
"masksPerSecond": 15.81690352996101,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 13.626832980662584,
"totalMs": 4784.854124998674,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2547598485541728,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 5155.519834021106,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 905.0123339984566,
"inferenceMs": 989.9023750331253,
"masksPerSecond": 1.0102006270734898,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753820160,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 892.3573340289295,
"totalMs": 6145.422209054232,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.2437045319363104,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 5155.519834021106,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 49.85554196173325,
"inferenceMs": 96.35416598757729,
"masksPerSecond": 10.378378451523597,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 40.81166599644348,
"totalMs": 5251.874000008684,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.25466691068814057,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4677.31666599866,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 911.1846670275554,
"inferenceMs": 991.0940420231782,
"masksPerSecond": 1.0089859867976216,
"imageEmbeddingReused": false,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753820160,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 899.5581250055693,
"totalMs": 5668.410708021838,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.24240197199853586,
"minimumForegroundRatio": 0.15,
"maximumForegroundRatio": 0.4,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4677.31666599866,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 56.050124985631555,
"inferenceMs": 104.56416697707027,
"masksPerSecond": 9.563505634002599,
"imageEmbeddingReused": true,
"inputSha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761373184,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 48.713708005379885,
"totalMs": 4781.88083297573,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4566.280416969676,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 530.0289170118049,
"inferenceMs": 582.4900830048136,
"masksPerSecond": 1.7167674251919172,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4584308736,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 499.31720801396295,
"totalMs": 5148.770499974489,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4566.280416969676,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 47.90825000964105,
"inferenceMs": 77.20254198648036,
"masksPerSecond": 12.952941370442428,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 4593958912,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 36.93633299553767,
"totalMs": 4643.482958956156,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4533.564290963113,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 888.4584159823135,
"inferenceMs": 939.1801250167191,
"masksPerSecond": 1.0647584774882222,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 877.2589580039494,
"totalMs": 5472.744415979832,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4533.564290963113,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 23.463624995201826,
"inferenceMs": 52.2713340469636,
"masksPerSecond": 19.13094467995674,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 16.233125003054738,
"totalMs": 4585.835625010077,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 2,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37089873731121564,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4408.360000001267,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 586.7768339812756,
"inferenceMs": 637.5496670370921,
"masksPerSecond": 1.5685052501828392,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 578.6690840031952,
"totalMs": 5045.909667038359,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp32",
"device": "cpu",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3677469670710572,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4408.360000001267,
"device": "cpu",
"provider": "PyTorch CPU",
"acceleration": "cpu",
"placement": {
"cpu/torch.float32": 314
},
"parameterDtype": "torch.float32",
"loadingInfo": {
"missing_keys": [],
"unexpected_keys": [],
"mismatched_keys": [],
"error_msgs": []
},
"accelerationDetails": "Image encoder and prompt/mask decoder on selected device; image preparation, mask resizing/binarization and serialization on CPU. One-image embeddings retained until replacement/disposal.",
"durationMs": 18.08595797047019,
"inferenceMs": 47.1801669918932,
"masksPerSecond": 21.19534676873497,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"modelInferenceMs": 13.159875001292676,
"totalMs": 4455.54016699316,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "cpu",
"actualDevice": "cpu",
"requestedPrecision": "fp32",
"actualPrecision": "fp32",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.37081703391928694,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4678.133665991481,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 502.2778330021538,
"inferenceMs": 555.0298329908401,
"masksPerSecond": 1.8017049545091812,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753836544,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 489.9972919956781,
"totalMs": 5233.1634989823215,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "fp16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3678583807873236,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4678.133665991481,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.float16": 314
},
"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": 51.40366597333923,
"inferenceMs": 81.0687500052154,
"masksPerSecond": 12.335209312289468,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761389568,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 43.663958029355854,
"totalMs": 4759.202415996697,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "fp16",
"actualPrecision": "fp16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.3708541718247091,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4581.378458009567,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 506.27783295931295,
"inferenceMs": 558.2267080317251,
"masksPerSecond": 1.7913868785066587,
"imageEmbeddingReused": false,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3753836544,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 494.6050000144169,
"totalMs": 5139.605166041292,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "good"
}
},
{
"precision": "bf16",
"device": "mps",
"status": "pass",
"quality": {
"passed": true,
"metric": "independent foreground/background probe accuracy and whole-subject area bounds",
"score": 1,
"minimum": 0.9,
"foregroundRatio": 0.36806635305768753,
"minimumForegroundRatio": 0.25,
"maximumForegroundRatio": 0.8,
"promptSemanticsPassed": true,
"speedPassed": true
},
"runtime": {
"engine": "transformers",
"version": "5.22.1",
"torchVersion": "2.10.0",
"loadMs": 4581.378458009567,
"device": "mps",
"provider": "Metal/MPS",
"acceleration": "partial",
"placement": {
"mps:0/torch.bfloat16": 314
},
"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": 55.95825001364574,
"inferenceMs": 87.99379202537239,
"masksPerSecond": 11.364438069809028,
"imageEmbeddingReused": true,
"inputSha256": "d4cb12ea8a768582562e1f1c0c930ea6fc4757f5b66551b986b41498efd75ca1",
"processedShape": [
1,
3,
1024,
1024
],
"driverAllocatedMemoryBytes": 3761389568,
"memoryMeasurementScope": "Post-inference driver allocation including one cached image embedding, not peak",
"modelInferenceMs": 47.371333988849074,
"totalMs": 4669.3722500349395,
"artifactRevision": "d67c795ed0caaec7423d799da25cdeded48aac40",
"requestedDevice": "mps",
"actualDevice": "mps",
"requestedPrecision": "bf16",
"actualPrecision": "bf16",
"cpuThreads": 4,
"fallback": false,
"grade": "excellent"
}
}
]Release validation still required
- Sample publication excluded/pending
- Complete model/runtime backup excluded/pending
- Forced model-source and runtime-source recovery excluded/pending