A model is still running

A model is still running or downloading. Leaving will stop the current operation. Stay on this page to let it finish.

Skip to content
LocalAI.toolsBeta

Join the waitlist

Get an email when the new apps and model runs launch.

No account needed.

LaMa

Remove masked image regions locally with Big LaMa. Supply an image and matching mask: white removes, black keeps. Returns a same-size PNG with unmasked pixels preserved.

Desktop inferencePrivate · Runs on your computer
About this model

Remove masked image regions locally with Big LaMa. Supply an image and matching mask: white removes, black keeps. Returns a same-size PNG with unmasked pixels preserved.

Created by
LaMa authors
License
Apache-2.0
Model released
Unknown
Source repository created
Aug 30, 2021
Model weights · FP32
205.7 MB

Best for

  • Removing unwanted objects or marks with a painted mask
  • Local image restoration experiments with reviewed output

Keep in mind

  • Generated regions are plausible guesses and may contain distorted textures, edges or objects; not recovered ground truth or evidence.
  • Large masks, faces, text, complex structure and poor input quality can fail; review all generated content before use.
  • Accepts opaque still PNG/JPEG and a matching oriented PNG/JPEG mask; grayscale values above127 remove. Transparency, animation, HDR and video are unsupported.
  • No fixed image dimension/file-size cap; full-image FFT/activations require increasing memory. Available host memory bounds practical resolution.
  • Current measured quality scope is independent authored gradient/texture and a cleared natural-photo small-mark removal, irregular dimensions and empty masks, not broad natural-image quality certification.
  • Only FP32 CPU/Metal on tested Mac verified. Other precision/platform/hardware combinations remain unsupported or unverified.
  • The app downloads its own pinned CPython and isolated packages on request; installed Python, user packages and Python/pip configuration are not used. Automatic model-package provisioning is verified on Apple Silicon Mac; other platform stacks remain unverified.

Processed on your deviceYour input is processed locally in the desktop app. Model and runtime files download from external hosts.

Target platforms · verification required

  • macos · aarch64, x86_64
  • windows · x86_64

Recommended configuration

FP32 · mps (verified platform only)

Sources & technical details
Original model
https://github.com/advimman/lama
Release verification
Unknown
Release evidence
Exact original checkpoint public release date has not been verified. Repository creation, training, paper, artifact conversion and update dates are not release evidence.
Source updated
Feb 5, 2025
Native model files
https://huggingface.co/signature-ai/big-lama
Files updated
Aug 2, 2024
Artifact revision
0c69ed801efa8bda408da0a9203da9800d0ea9e7

Desktop app

Sample result

LaMa — Native Image Inpainting

Run LaMa 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

macOS Apple SiliconCPU / FP32Tested8.0 GiBRAM estimate
macOS Apple SiliconMPS / FP32Tested8.0 GiBRAM estimate
Windows Intel / AMDCPU / FP32Unverified8.0 GiBRAM estimate
macOS Intel / AMDCPU / FP32Unverified8.0 GiBRAM estimate
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.

LaMa — Native Image Inpainting

31 configurations tested
Passed31
Failed0
TestedNov 1, 2026
Runtimepytorch 2.10.0

Quality criteria

No explicit quality gate recorded. Review per-case evidence; full certification is pending.

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6432 ms / 0 ms2.020 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6433 ms / 0 ms2.016 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6433 ms / 0 ms2.006 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6433 ms / 0 ms2.005 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6433 ms / 0 ms2.002 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6433 ms / 0 ms1.978 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64100 ms / 0 ms0.498 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64138 ms / 0 ms0.475 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64142 ms / 0 ms0.461 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64209 ms / 0 ms0.471 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64233 ms / 0 ms0.522 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64354 ms / 394 ms0.185 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64368 ms / 421 ms0.178 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64378 ms / 395 ms0.174 MP/s
FP32cpuPassedFastmacos · aarch64144 ms / 0 ms0.347 MP/s
FP32cpuPassedFastmacos · aarch64197 ms / 0 ms0.332 MP/s
FP32cpuPassedFastmacos · aarch64199 ms / 0 ms0.329 MP/s
FP32cpuPassedFastmacos · aarch64202 ms / 0 ms0.325 MP/s
FP32cpuPassedFastmacos · aarch64202 ms / 0 ms0.325 MP/s
FP32cpuPassedFastmacos · aarch64206 ms / 0 ms0.319 MP/s
FP32cpuPassedFastmacos · aarch64209 ms / 0 ms0.314 MP/s
FP32cpuPassedFastmacos · aarch64229 ms / 0 ms0.287 MP/s
FP32cpuPassedFastmacos · aarch64304 ms / 0 ms0.215 MP/s
FP32cpuPassedFastmacos · aarch64307 ms / 0 ms0.214 MP/s
FP32cpuPassedFastmacos · aarch64321 ms / 0 ms0.379 MP/s
FP32cpuPassedFastmacos · aarch64344 ms / 0 ms0.190 MP/s
FP32cpuPassedFastmacos · aarch64404 ms / 0 ms0.243 MP/s
FP32cpuPassedFastmacos · aarch64437 ms / 194 ms0.150 MP/s
FP32cpuPassedFastmacos · aarch64438 ms / 189 ms0.150 MP/s
FP32cpuPassedFastmacos · aarch64448 ms / 198 ms0.146 MP/s
FP32cpuPassedFastmacos · aarch64482 ms / 186 ms0.136 MP/s
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-01T18:52:33.959686+00:00
Environment
{ "hardware": "Apple M3 Max", "os": "macOS-27.2-arm64-arm-64bit-Mach-O", "python": "3.13.1", "runtime": "Torch2.10.0/Numpy2.3.4/Pillow12.0.0" }
Runtime
pytorch 2.10.0
Artifact revision
0c69ed801efa8bda408da0a9203da9800d0ea9e7
Fixture
{
  "file": "model-samples/desktop/lama/gradient-input.png",
  "size": 1037,
  "sha1": "e7dcf844adedc8ccdc6a73e8c777416a36df23f0",
  "sha256": "54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51",
  "url": "https://assets.localai.tools/samples/v1/files/54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51/gradient-input.png"
}
Rerun when
artifact revision/checksum; Torch/NumPy/Pillow version or OS/driver/device; preprocessing/mask threshold/padding/compositing; fixtures/independent references/quality gates
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 186.38020799699007,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 487.87658299988834,
      "inferenceMs": 481.85020799792255,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.13600907276205337,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 674.2567909968784,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 346.47729200150934,
      "inferenceMs": 344.2481249985576,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.1903743121339429,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 346.47729200150934,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 230.7752079977945,
      "inferenceMs": 228.50883399951272,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.2867985401393267,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 230.7752079977945,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 197.87833300142665,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 453.92029199865647,
      "inferenceMs": 447.8572500011069,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.14633234138743545,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 651.7986250000831,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.7830555555555555,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9910600741452524,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 411.4829579993966,
      "inferenceMs": 404.36104200125556,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.2431094734385781,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 411.4829579993966,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.8472222222222222,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9894449058693244,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 145.8755830026348,
      "inferenceMs": 143.89187499909895,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.34739279059580686,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 145.8755830026348,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 210.66462499948102,
      "inferenceMs": 208.57375000196043,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.31421020142459927,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 210.66462499948102,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.828125,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8096214432164826,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 331.86033299716655,
      "inferenceMs": 321.08983299985994,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.37910885829902036,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 331.86033299716655,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 207.85729099952732,
      "inferenceMs": 205.50208400163683,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3189067415952726,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 207.85729099952732,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 203.7262500016368,
      "inferenceMs": 201.62516700293054,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.32503878843184025,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 203.7262500016368,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 203.9706670002488,
      "inferenceMs": 201.7993330009631,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3247582587385818,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 203.9706670002488,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 193.9359579992015,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 443.3363330026623,
      "inferenceMs": 437.0553749977262,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.14994896241772784,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 637.2722910018638,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 306.35833400083357,
      "inferenceMs": 304.2964169981133,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.215368950599265,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 306.35833400083357,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 199.40595800289884,
      "inferenceMs": 197.38283400147338,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3320248203524669,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 199.40595800289884,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 189.08400000145775,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 444.055957999808,
      "inferenceMs": 438.24591700104065,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.14954160999027488,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 633.1399580012658,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 308.7557079998078,
      "inferenceMs": 306.5203329970245,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.21380637088318763,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 308.7557079998078,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 201.58816699768067,
      "inferenceMs": 199.45820800057845,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.32857008321166675,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 201.58816699768067,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 420.8050410015858,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 383.67837500118185,
      "inferenceMs": 367.5656249979511,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.17829741287794612,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1354776576,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 804.4834160027676,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.7830555555555555,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9910600741452524,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 222.61349999826052,
      "inferenceMs": 208.6115840029379,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.471229823932575,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1381695488,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 222.61349999826052,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.8472222222222222,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9894449058693244,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 109.3805419986893,
      "inferenceMs": 100.42629200324882,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.4977481394850554,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "driverAllocatedMemoryBytes": 1382891520,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 109.3805419986893,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 35.69195800082525,
      "inferenceMs": 32.74341600263142,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.001501614698149,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1382891520,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.69195800082525,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.828125,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8096214432164826,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 252.90804199903505,
      "inferenceMs": 233.19008300313726,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.52201190733468,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "driverAllocatedMemoryBytes": 1424769024,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 252.90804199903505,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 37.16875000100117,
      "inferenceMs": 33.13120800157776,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 1.9780745693570565,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1424769024,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.16875000100117,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 36.00250000090455,
      "inferenceMs": 32.502792000741465,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.016319090326301,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1424769024,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.00250000090455,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 35.35787500004517,
      "inferenceMs": 32.44733300016378,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.019765384097029,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1424769024,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.35787500004517,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 394.9090409987548,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 394.8847499996191,
      "inferenceMs": 377.67075000010664,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.17352680873480802,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1354776576,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 789.7937909983739,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 147.48291700016125,
      "inferenceMs": 142.11720900129876,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.4611404942479632,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1354776576,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 147.48291700016125,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 36.47254099996644,
      "inferenceMs": 32.67783399860491,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.0055184809004745,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1354776576,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.47254099996644,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 393.58891699885135,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 370.48562500058324,
      "inferenceMs": 353.78762500113226,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.1852410750652747,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1354776576,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 764.0745419994346,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 141.4737500017509,
      "inferenceMs": 137.95362500241026,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.47505819436680247,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1354776576,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 141.4737500017509,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0.9411111111111111,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9881947106170946,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 36.03687500071828,
      "inferenceMs": 32.68779200152494,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.0049075201207422,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1354776576,
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.03687500071828,
      "artifactRevision": "0c69ed801efa8bda408da0a9203da9800d0ea9e7",
      "fallback": false,
      "grade": "fast"
    }
  }
]

LaMa · CelebA-HQ Fourier

24 configurations tested
Passed24
Failed0
TestedNov 9, 2026
Runtimepytorch 2.10.0

Quality criteria

No explicit quality gate recorded. Review per-case evidence; full certification is pending.

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6422 ms / 0 ms2.928 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6424 ms / 0 ms2.772 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6425 ms / 0 ms2.629 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6430 ms / 0 ms2.215 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6493 ms / 0 ms0.540 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6493 ms / 0 ms0.538 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6497 ms / 0 ms1.258 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch64100 ms / 0 ms1.219 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64145 ms / 0 ms0.679 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64147 ms / 0 ms0.667 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64224 ms / 497 ms0.293 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64235 ms / 518 ms0.279 MP/s
FP32cpuPassedVery fastmacos · aarch6490 ms / 0 ms0.554 MP/s
FP32cpuPassedVery fastmacos · aarch6494 ms / 0 ms0.534 MP/s
FP32cpuPassedFastmacos · aarch64145 ms / 0 ms0.453 MP/s
FP32cpuPassedFastmacos · aarch64146 ms / 0 ms0.449 MP/s
FP32cpuPassedFastmacos · aarch64147 ms / 0 ms0.445 MP/s
FP32cpuPassedFastmacos · aarch64150 ms / 0 ms0.437 MP/s
FP32cpuPassedFastmacos · aarch64217 ms / 0 ms0.562 MP/s
FP32cpuPassedFastmacos · aarch64220 ms / 0 ms0.554 MP/s
FP32cpuPassedFastmacos · aarch64243 ms / 381 ms0.270 MP/s
FP32cpuPassedFastmacos · aarch64244 ms / 425 ms0.268 MP/s
FP32cpuPassedFastmacos · aarch64247 ms / 0 ms0.397 MP/s
FP32cpuPassedFastmacos · aarch64253 ms / 0 ms0.388 MP/s
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-09T17:58:18.442527+00:00
Environment
{ "os": "macOS-27.2-arm64-arm-64bit-Mach-O", "hardware": "Apple M3 Max", "python": "3.13.1", "runtimeFile": "research/investigation-completion/harnesses/lama-candidate-runtime-4b4a957385202b38e88930775debcdf78937d9c57c3a676598bcced3846e8511.py", "runtimeSha256": "4b4a957385202b38e88930775debcdf78937d9c57c3a676598bcced3846e8511", "cpuThreads": 4, "implicitMpsFallback": false, "originalRuntimeFile": "localai-desktop/crates/desktopai/src/models/lama/runtime.py" }
Runtime
pytorch 2.10.0
Artifact revision
3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281
Fixture
{
  "file": "model-samples/desktop/lama/gradient-input.png",
  "size": 1037,
  "sha1": "e7dcf844adedc8ccdc6a73e8c777416a36df23f0",
  "sha256": "54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51",
  "url": "https://assets.localai.tools/samples/v1/files/54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51/gradient-input.png"
}
Rerun when
artifact revision/checksum; Torch/NumPy/Pillow version or OS/driver/device; preprocessing/mask threshold/padding/compositing; fixtures/independent references/quality gates
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.5833333333333335,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9675946897104428,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 424.8522079869872,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 250.62958300986793,
      "inferenceMs": 244.2417920101434,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.2683242677701868,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 675.4817909968551,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.0775,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9762817646443216,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 254.55912500910927,
      "inferenceMs": 247.4385420064209,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.39728653104272277,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "totalMs": 254.55912500910927,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 3.1666666666666665,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9605481727574751,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 95.57004200178199,
      "inferenceMs": 93.58241700101644,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.5341494866440248,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "totalMs": 95.57004200178199,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 152.0873749977909,
      "inferenceMs": 149.98437500617,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.4369521825010372,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 152.0873749977909,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.9,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8089313397009552,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 230.37020899937488,
      "inferenceMs": 219.8401250061579,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.5537114755397373,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "totalMs": 230.37020899937488,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.5833333333333335,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9675946897104428,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 148.21941699483432,
      "inferenceMs": 145.883833989501,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.44923414889628216,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 148.21941699483432,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.5833333333333335,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9675946897104428,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 380.72833299520425,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 248.90112499997485,
      "inferenceMs": 243.0705000006128,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.26961725096148964,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 629.6294579951791,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.0775,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9762817646443216,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 259.9596660002135,
      "inferenceMs": 253.13066599483136,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.3883527885238814,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "totalMs": 259.9596660002135,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 3.1666666666666665,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9605481727574751,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 92.10912499111146,
      "inferenceMs": 90.16670800338034,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.5543842190415335,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "totalMs": 92.10912499111146,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 146.69662500091363,
      "inferenceMs": 144.56149999750778,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.45334338673249674,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 146.69662500091363,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.9,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8089313397009552,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 227.49216700321995,
      "inferenceMs": 216.59749999525957,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.5620009464682839,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "totalMs": 227.49216700321995,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.5833333333333335,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9675946897104428,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 149.533584000892,
      "inferenceMs": 147.37462499761023,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.4446898507871535,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 149.533584000892,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.5833333333333335,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9675946897104428,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 517.7525420003803,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 252.5998330092989,
      "inferenceMs": 234.94449999998324,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.27894247364805164,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1254113280,
      "totalMs": 770.3523750096792,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.0775,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9762817646443216,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 158.9477910019923,
      "inferenceMs": 144.81470800819807,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.6788260761084774,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1281032192,
      "totalMs": 158.9477910019923,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 3.1666666666666665,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9605481727574751,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 103.60941699764226,
      "inferenceMs": 92.97383300145157,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.5376458987037737,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "driverAllocatedMemoryBytes": 1282228224,
      "totalMs": 103.60941699764226,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 33.24320800311398,
      "inferenceMs": 29.587875003926456,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.2149613647922686,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1282228224,
      "totalMs": 33.24320800311398,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.9,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8089313397009552,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 117.97583299630787,
      "inferenceMs": 99.89733400288969,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 1.2185310170187207,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "driverAllocatedMemoryBytes": 1324105728,
      "totalMs": 117.97583299630787,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.5833333333333335,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9675946897104428,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 27.562167000724003,
      "inferenceMs": 23.64287499221973,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.7719133151770343,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1324105728,
      "totalMs": 27.562167000724003,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.5833333333333335,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9675946897104428,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 497.04345800273586,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 241.48087500361726,
      "inferenceMs": 224.03758400469087,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.29252234749428385,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1254113280,
      "totalMs": 738.5243330063531,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.0775,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9762817646443216,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 161.730791995069,
      "inferenceMs": 147.39383300184272,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.6669478498382697,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1281032192,
      "totalMs": 161.730791995069,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 3.1666666666666665,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9605481727574751,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 102.53370799182449,
      "inferenceMs": 92.56683300191071,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.5400098326683402,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "driverAllocatedMemoryBytes": 1282228224,
      "totalMs": 102.53370799182449,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 28.77349998743739,
      "inferenceMs": 24.930500003392808,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.6287479188576706,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1282228224,
      "totalMs": 28.77349998743739,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.9,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8089313397009552,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 115.14337500557303,
      "inferenceMs": 96.7885840072995,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 1.2576689828504946,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "driverAllocatedMemoryBytes": 1324105728,
      "totalMs": 115.14337500557303,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.5833333333333335,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9675946897104428,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 26.30204199522268,
      "inferenceMs": 22.38362499338109,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.9278546267362504,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1324105728,
      "totalMs": 26.30204199522268,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "3060742ea7464288d7d5fa8e83c940ddbd359ba1640f79135759da9c8c81f281",
      "cpuThreads": 4
    }
  }
]

LaMa · CelebA-HQ Fourier Shallow

24 configurations tested
Passed24
Failed0
TestedNov 9, 2026
Runtimepytorch 2.10.0

Quality criteria

No explicit quality gate recorded. Review per-case evidence; full certification is pending.

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6419 ms / 0 ms3.512 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6419 ms / 0 ms3.475 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6425 ms / 0 ms2.654 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6430 ms / 0 ms2.219 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6489 ms / 0 ms1.374 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6490 ms / 0 ms0.558 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6492 ms / 0 ms0.541 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6493 ms / 0 ms1.303 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64132 ms / 0 ms0.745 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64141 ms / 0 ms0.699 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64192 ms / 460 ms0.341 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64194 ms / 452 ms0.337 MP/s
FP32cpuPassedVery fastmacos · aarch6471 ms / 0 ms0.702 MP/s
FP32cpuPassedVery fastmacos · aarch6472 ms / 0 ms0.697 MP/s
FP32cpuPassedFastmacos · aarch64120 ms / 0 ms0.548 MP/s
FP32cpuPassedFastmacos · aarch64123 ms / 0 ms0.533 MP/s
FP32cpuPassedFastmacos · aarch64124 ms / 0 ms0.530 MP/s
FP32cpuPassedFastmacos · aarch64127 ms / 0 ms0.518 MP/s
FP32cpuPassedFastmacos · aarch64185 ms / 0 ms0.659 MP/s
FP32cpuPassedFastmacos · aarch64186 ms / 0 ms0.654 MP/s
FP32cpuPassedFastmacos · aarch64193 ms / 394 ms0.339 MP/s
FP32cpuPassedFastmacos · aarch64212 ms / 0 ms0.464 MP/s
FP32cpuPassedFastmacos · aarch64212 ms / 0 ms0.464 MP/s
FP32cpuPassedFastmacos · aarch64227 ms / 375 ms0.289 MP/s
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-09T17:58:18.442527+00:00
Environment
{ "os": "macOS-27.2-arm64-arm-64bit-Mach-O", "hardware": "Apple M3 Max", "python": "3.13.1", "runtimeFile": "research/investigation-completion/harnesses/lama-candidate-runtime-4b4a957385202b38e88930775debcdf78937d9c57c3a676598bcced3846e8511.py", "runtimeSha256": "4b4a957385202b38e88930775debcdf78937d9c57c3a676598bcced3846e8511", "cpuThreads": 4, "implicitMpsFallback": false, "originalRuntimeFile": "localai-desktop/crates/desktopai/src/models/lama/runtime.py" }
Runtime
pytorch 2.10.0
Artifact revision
5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06
Fixture
{
  "file": "model-samples/desktop/lama/gradient-input.png",
  "size": 1037,
  "sha1": "e7dcf844adedc8ccdc6a73e8c777416a36df23f0",
  "sha256": "54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51",
  "url": "https://assets.localai.tools/samples/v1/files/54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51/gradient-input.png"
}
Rerun when
artifact revision/checksum; Torch/NumPy/Pillow version or OS/driver/device; preprocessing/mask threshold/padding/compositing; fixtures/independent references/quality gates
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.9747222222222223,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9626851109794766,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 394.3503329937812,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 199.92108299629763,
      "inferenceMs": 193.42645800497849,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3388161096260844,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 594.2714159900788,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 3.8730555555555557,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9557824100061206,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 218.7306670093676,
      "inferenceMs": 211.68295800453052,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.46439260357414347,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "totalMs": 218.7306670093676,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.3880555555555554,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9702484772978959,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 73.7316250015283,
      "inferenceMs": 71.7529589892365,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.6966541966234233,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "totalMs": 73.7316250015283,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 125.0720829993952,
      "inferenceMs": 122.95529199764132,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.5330067452587335,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 125.0720829993952,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.652083333333334,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8113116967545132,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 197.31370900990441,
      "inferenceMs": 186.23837499762885,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.6536139504092527,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "totalMs": 197.31370900990441,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.9747222222222223,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9626851109794766,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 125.90800000180025,
      "inferenceMs": 123.73037499492057,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.5296678362341536,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 125.90800000180025,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.9747222222222223,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9626851109794766,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 375.2985409955727,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 236.37170800066087,
      "inferenceMs": 226.6457499936223,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.2891560949271899,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 611.6702489962336,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 3.8730555555555557,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9557824100061206,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 218.9467499993043,
      "inferenceMs": 212.0356250088662,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.4636202053116756,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "totalMs": 218.9467499993043,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.3880555555555554,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9702484772978959,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 73.07708299777005,
      "inferenceMs": 71.19054198847152,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.7021578794567235,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "totalMs": 73.07708299777005,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 121.68583300081082,
      "inferenceMs": 119.57116700068582,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.5480919994669292,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 121.68583300081082,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.652083333333334,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8113116967545132,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 195.69916700129397,
      "inferenceMs": 184.8244159918977,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.6586142818129413,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "totalMs": 195.69916700129397,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.9747222222222223,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9626851109794766,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 128.90574999619275,
      "inferenceMs": 126.61941700207535,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.5175825442232594,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 128.90574999619275,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.9747222222222223,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9626851109794766,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 452.20216699817684,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 213.27704199939035,
      "inferenceMs": 194.3440000031842,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3372164821086642,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1212170240,
      "totalMs": 665.4792089975672,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 3.8730555555555557,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9557824100061206,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 147.3009580076905,
      "inferenceMs": 131.94729200040456,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.745024763370654,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1239089152,
      "totalMs": 147.3009580076905,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.3880555555555554,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9702484772978959,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 103.2462500006659,
      "inferenceMs": 92.45933299825992,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.5406376877166175,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "driverAllocatedMemoryBytes": 1240285184,
      "totalMs": 103.2462500006659,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 32.817332990816794,
      "inferenceMs": 29.53983299084939,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.218563660136508,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1240285184,
      "totalMs": 32.817332990816794,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.652083333333334,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8113116967545132,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 111.34729199693538,
      "inferenceMs": 93.40766700915992,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 1.303190668364117,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "driverAllocatedMemoryBytes": 1282179072,
      "totalMs": 111.34729199693538,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.9747222222222223,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9626851109794766,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 22.397000007913448,
      "inferenceMs": 18.86099998955615,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 3.4746832106616337,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1282162688,
      "totalMs": 22.397000007913448,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.9747222222222223,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9626851109794766,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 459.9296670057811,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 211.15624999220017,
      "inferenceMs": 191.95100000069942,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.34142046668035697,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1212170240,
      "totalMs": 671.0859169979813,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 3.8730555555555557,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9557824100061206,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 155.02533299149945,
      "inferenceMs": 140.66975000605453,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.6988282839471096,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1239089152,
      "totalMs": 155.02533299149945,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.3880555555555554,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9702484772978959,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 98.88241700537037,
      "inferenceMs": 89.57666700007394,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.5580359447841338,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "driverAllocatedMemoryBytes": 1240285184,
      "totalMs": 98.88241700537037,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 28.593374998308718,
      "inferenceMs": 24.692791994311847,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.654053863779222,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1240285184,
      "totalMs": 28.593374998308718,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 19.652083333333334,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8113116967545132,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 106.93779098801315,
      "inferenceMs": 88.61137500207406,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 1.3737288242863945,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "driverAllocatedMemoryBytes": 1282179072,
      "totalMs": 106.93779098801315,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.9747222222222223,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9626851109794766,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 21.841124995262362,
      "inferenceMs": 18.66225000412669,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 3.5116880325527924,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1282179072,
      "totalMs": 21.841124995262362,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "5ba34746ca7e365c56429317e64c6ff964ecb1217b0599c735a76118197cbb06",
      "cpuThreads": 4
    }
  }
]

LaMa · Places Dilated

24 configurations tested
Passed24
Failed0
TestedNov 9, 2026
Runtimepytorch 2.10.0

Quality criteria

No explicit quality gate recorded. Review per-case evidence; full certification is pending.

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6435 ms / 0 ms1.868 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6435 ms / 0 ms1.866 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms1.845 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms1.835 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6480 ms / 0 ms0.627 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6480 ms / 0 ms0.624 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6491 ms / 0 ms1.075 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6492 ms / 0 ms1.064 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64103 ms / 0 ms1.180 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64107 ms / 0 ms1.138 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64152 ms / 1315 ms0.431 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64155 ms / 1343 ms0.422 MP/s
FP32cpuPassedFastmacos · aarch64115 ms / 0 ms0.434 MP/s
FP32cpuPassedFastmacos · aarch64115 ms / 0 ms0.433 MP/s
FP32cpuPassedFastmacos · aarch64165 ms / 0 ms0.397 MP/s
FP32cpuPassedFastmacos · aarch64165 ms / 0 ms0.397 MP/s
FP32cpuPassedFastmacos · aarch64166 ms / 0 ms0.394 MP/s
FP32cpuPassedFastmacos · aarch64167 ms / 0 ms0.392 MP/s
FP32cpuPassedFastmacos · aarch64184 ms / 1279 ms0.357 MP/s
FP32cpuPassedFastmacos · aarch64186 ms / 1273 ms0.353 MP/s
FP32cpuPassedFastmacos · aarch64240 ms / 0 ms0.410 MP/s
FP32cpuPassedFastmacos · aarch64241 ms / 0 ms0.408 MP/s
FP32cpuPassedFastmacos · aarch64274 ms / 0 ms0.444 MP/s
FP32cpuPassedFastmacos · aarch64275 ms / 0 ms0.443 MP/s
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-09T17:58:18.442527+00:00
Environment
{ "os": "macOS-27.2-arm64-arm-64bit-Mach-O", "hardware": "Apple M3 Max", "python": "3.13.1", "runtimeFile": "research/investigation-completion/harnesses/lama-candidate-runtime-4b4a957385202b38e88930775debcdf78937d9c57c3a676598bcced3846e8511.py", "runtimeSha256": "4b4a957385202b38e88930775debcdf78937d9c57c3a676598bcced3846e8511", "cpuThreads": 4, "implicitMpsFallback": false, "originalRuntimeFile": "localai-desktop/crates/desktopai/src/models/lama/runtime.py" }
Runtime
pytorch 2.10.0
Artifact revision
e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0
Fixture
{
  "file": "model-samples/desktop/lama/gradient-input.png",
  "size": 1037,
  "sha1": "e7dcf844adedc8ccdc6a73e8c777416a36df23f0",
  "sha256": "54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51",
  "url": "https://assets.localai.tools/samples/v1/files/54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51/gradient-input.png"
}
Rerun when
artifact revision/checksum; Torch/NumPy/Pillow version or OS/driver/device; preprocessing/mask threshold/padding/compositing; fixtures/independent references/quality gates
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.2258333333333333,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9846231576013101,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 1279.148499990697,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 189.65729199408088,
      "inferenceMs": 183.59050000435673,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3569683616442288,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 1468.805791984778,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.4577777777777778,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9833569596006685,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 247.1202079905197,
      "inferenceMs": 240.00745901139453,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.4095872703495142,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "totalMs": 247.1202079905197,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.4233333333333333,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9822674418604651,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 117.39458299416583,
      "inferenceMs": 115.41366700839717,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.43311161750335064,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "totalMs": 117.39458299416583,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 167.07275000226218,
      "inferenceMs": 165.00183299649507,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.39718346644907937,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 167.07275000226218,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 21.380208333333332,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.7947192078811822,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 285.03129100135993,
      "inferenceMs": 274.1468750027707,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.44402475862170504,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "totalMs": 285.03129100135993,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.2258333333333333,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9846231576013101,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 168.59675000887364,
      "inferenceMs": 166.41533400979824,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3938098636760321,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 168.59675000887364,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.2258333333333333,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9846231576013101,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 1273.2218750024913,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 192.14258299325593,
      "inferenceMs": 185.7209169975249,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3528735538220145,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 1465.3644579957472,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.4577777777777778,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9833569596006685,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 247.67954199342057,
      "inferenceMs": 240.76491700543556,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.4082986891224716,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "totalMs": 247.67954199342057,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.4233333333333333,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9822674418604651,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 117.11058299988508,
      "inferenceMs": 115.1653749984689,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.43404538908213136,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "totalMs": 117.11058299988508,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 169.42504100734368,
      "inferenceMs": 167.28437499841675,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3917640245875938,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 169.42504100734368,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 21.380208333333332,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.7947192078811822,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 286.12187500402797,
      "inferenceMs": 274.81808399898,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.44294028336378255,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "totalMs": 286.12187500402797,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.2258333333333333,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9846231576013101,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 167.15808300068602,
      "inferenceMs": 165.01083299226593,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.3971618033288255,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 167.15808300068602,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.2258333333333333,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9846231576013101,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 1343.2003339985386,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 175.43908400693908,
      "inferenceMs": 155.22841599886306,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.4221907411622367,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1321074688,
      "totalMs": 1518.6394180054776,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.4577777777777778,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9833569596006685,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 105.6110829958925,
      "inferenceMs": 91.41645800264087,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 1.075342472765245,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1347846144,
      "totalMs": 105.6110829958925,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.4233333333333333,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9822674418604651,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 90.1335420057876,
      "inferenceMs": 80.16791599220596,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.6235287444026836,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "driverAllocatedMemoryBytes": 1348894720,
      "totalMs": 90.1335420057876,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 38.87895800289698,
      "inferenceMs": 35.716290993150324,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 1.834904974107432,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1348894720,
      "totalMs": 38.87895800289698,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 21.380208333333332,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.7947192078811822,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 123.2435000129044,
      "inferenceMs": 103.13854200649075,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 1.1802377426698487,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "driverAllocatedMemoryBytes": 1382236160,
      "totalMs": 123.2435000129044,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.2258333333333333,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9846231576013101,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 38.38608298974577,
      "inferenceMs": 35.11579199403059,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 1.8662828396734041,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1382236160,
      "totalMs": 38.38608298974577,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.2258333333333333,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9846231576013101,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 1314.962500007823,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 168.73695800313726,
      "inferenceMs": 151.95637500437442,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.4312816754026502,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1321074688,
      "totalMs": 1483.6994580109604,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.4577777777777778,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9833569596006685,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 108.78641600720584,
      "inferenceMs": 92.36883399717044,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 1.0642550711748875,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1347846144,
      "totalMs": 108.78641600720584,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.4233333333333333,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9822674418604651,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 88.96108399494551,
      "inferenceMs": 79.77579200814944,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.6265935911346842,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "driverAllocatedMemoryBytes": 1348894720,
      "totalMs": 88.96108399494551,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 39.44454199518077,
      "inferenceMs": 35.5163329950301,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 1.8452355430153962,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1348894720,
      "totalMs": 39.44454199518077,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 21.380208333333332,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.7947192078811822,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 127.52654201176483,
      "inferenceMs": 106.92608400131576,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 1.1384312923916873,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "driverAllocatedMemoryBytes": 1382236160,
      "totalMs": 127.52654201176483,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.2258333333333333,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9846231576013101,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire dilated-convolution graph on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 38.97062499891035,
      "inferenceMs": 35.07929199258797,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 1.8682247068682953,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1382236160,
      "totalMs": 38.97062499891035,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "e09dfa9ab3c74149a4ee5db115f23904acc6349f55f16972d11aea903c9496a0",
      "cpuThreads": 4
    }
  }
]

LaMa · Places Fourier

24 configurations tested
Passed24
Failed0
TestedNov 9, 2026
Runtimepytorch 2.10.0

Quality criteria

No explicit quality gate recorded. Review per-case evidence; full certification is pending.

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6422 ms / 0 ms2.914 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6423 ms / 0 ms2.912 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6423 ms / 0 ms2.882 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6423 ms / 0 ms2.814 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6495 ms / 0 ms0.529 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64101 ms / 0 ms1.201 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64107 ms / 0 ms1.137 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64111 ms / 0 ms0.451 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64162 ms / 0 ms0.606 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64167 ms / 0 ms0.588 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64233 ms / 507 ms0.281 MP/s
FP32mpsAcceleratedPassedFastmacos · aarch64317 ms / 521 ms0.207 MP/s
FP32cpuPassedVery fastmacos · aarch6486 ms / 0 ms0.581 MP/s
FP32cpuPassedVery fastmacos · aarch6489 ms / 0 ms0.560 MP/s
FP32cpuPassedFastmacos · aarch64145 ms / 0 ms0.453 MP/s
FP32cpuPassedFastmacos · aarch64146 ms / 0 ms0.450 MP/s
FP32cpuPassedFastmacos · aarch64146 ms / 0 ms0.448 MP/s
FP32cpuPassedFastmacos · aarch64154 ms / 0 ms0.425 MP/s
FP32cpuPassedFastmacos · aarch64237 ms / 0 ms0.513 MP/s
FP32cpuPassedFastmacos · aarch64240 ms / 0 ms0.507 MP/s
FP32cpuPassedFastmacos · aarch64248 ms / 391 ms0.264 MP/s
FP32cpuPassedFastmacos · aarch64249 ms / 0 ms0.396 MP/s
FP32cpuPassedFastmacos · aarch64267 ms / 441 ms0.246 MP/s
FP32cpuPassedFastmacos · aarch64273 ms / 0 ms0.360 MP/s
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-09T17:58:18.442527+00:00
Environment
{ "os": "macOS-27.2-arm64-arm-64bit-Mach-O", "hardware": "Apple M3 Max", "python": "3.13.1", "runtimeFile": "research/investigation-completion/harnesses/lama-candidate-runtime-4b4a957385202b38e88930775debcdf78937d9c57c3a676598bcced3846e8511.py", "runtimeSha256": "4b4a957385202b38e88930775debcdf78937d9c57c3a676598bcced3846e8511", "cpuThreads": 4, "implicitMpsFallback": false, "originalRuntimeFile": "localai-desktop/crates/desktopai/src/models/lama/runtime.py" }
Runtime
pytorch 2.10.0
Artifact revision
b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54
Fixture
{
  "file": "model-samples/desktop/lama/gradient-input.png",
  "size": 1037,
  "sha1": "e7dcf844adedc8ccdc6a73e8c777416a36df23f0",
  "sha256": "54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51",
  "url": "https://assets.localai.tools/samples/v1/files/54de72cfe485fe4ccf5eb987ce52329aa49a710e1a956bed3995e26942236e51/gradient-input.png"
}
Rerun when
artifact revision/checksum; Torch/NumPy/Pillow version or OS/driver/device; preprocessing/mask threshold/padding/compositing; fixtures/independent references/quality gates
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.2216666666666667,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9721314331509808,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 440.6410829978995,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 287.29491599369794,
      "inferenceMs": 266.58058298926335,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.24583936033570566,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 727.9359989915974,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.791111111111111,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9795513863386263,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 255.7147920015268,
      "inferenceMs": 248.50324999715667,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.3955843635893083,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "totalMs": 255.7147920015268,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.8716666666666666,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9766818936877076,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 89.37387500191107,
      "inferenceMs": 86.0276250023162,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.5810575381879268,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "totalMs": 89.37387500191107,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 146.7745410045609,
      "inferenceMs": 144.69483400171157,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.4529256379617864,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 146.7745410045609,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 18.8625,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8188928339250887,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 248.20758300484158,
      "inferenceMs": 237.27495899947826,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.5130250596746185,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "totalMs": 248.20758300484158,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.2216666666666667,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9721314331509808,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 147.97858300153166,
      "inferenceMs": 145.68120799958706,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.44985898249955314,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 147.97858300153166,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.2216666666666667,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9721314331509808,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 391.30641700467095,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 254.73179199616425,
      "inferenceMs": 248.31804099085275,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.2639196078484452,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 646.0382090008352,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.791111111111111,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9795513863386263,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 280.3120000025956,
      "inferenceMs": 273.05354199779686,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.36001730386194064,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "totalMs": 280.3120000025956,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.8716666666666666,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9766818936877076,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 91.60612500272691,
      "inferenceMs": 89.32070899754763,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.5596350562037347,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "totalMs": 91.60612500272691,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 156.57274999830406,
      "inferenceMs": 154.28550000069663,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.42477096032811956,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 156.57274999830406,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 18.8625,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8188928339250887,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 251.0831249965122,
      "inferenceMs": 239.89841600996442,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 0.5074147717379839,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "totalMs": 251.0831249965122,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.2216666666666667,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9721314331509808,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 148.399875004543,
      "inferenceMs": 146.14300000539515,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.44843748929186217,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "totalMs": 148.399875004543,
      "requestedDevice": "cpu",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.2216666666666667,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9721314331509808,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 520.526708001853,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 341.74295800039545,
      "inferenceMs": 317.06124999618623,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.20669823259950024,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1254113280,
      "totalMs": 862.2696660022484,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.791111111111111,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9795513863386263,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 181.5351249970263,
      "inferenceMs": 167.06354200141504,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.5884228169852125,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1281032192,
      "totalMs": 181.5351249970263,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.871388888888889,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9766853543743078,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 124.1117080062395,
      "inferenceMs": 110.7444579974981,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.4513724741072757,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "driverAllocatedMemoryBytes": 1282228224,
      "totalMs": 124.1117080062395,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 26.437708002049476,
      "inferenceMs": 22.4896669969894,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.914049372486175,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1282228224,
      "totalMs": 26.437708002049476,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 18.8625,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8188928339250887,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 119.99908401048742,
      "inferenceMs": 101.34550000657327,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 1.201118944522497,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "driverAllocatedMemoryBytes": 1324105728,
      "totalMs": 119.99908401048742,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.2216666666666667,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9721314331509808,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 26.655250010662712,
      "inferenceMs": 22.742624991224147,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.8816374550118478,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1324105728,
      "totalMs": 26.655250010662712,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.2216666666666667,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9721314331509808,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 507.1274590009125,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 251.46258299355395,
      "inferenceMs": 233.47125000145752,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 0.2807026561068691,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1254113280,
      "totalMs": 758.5900419944664,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.791111111111111,
      "maskedMaeBefore": 87.59083333333334,
      "maskedMaeImprovement": 0.9795513863386263,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 176.96412499935832,
      "inferenceMs": 162.2974579950096,
      "inputMegapixels": 0.098304,
      "megapixelsPerSecond": 0.6057026475610153,
      "maskedPixels": 1200,
      "paddedWidth": 384,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1281032192,
      "totalMs": 176.96412499935832,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 1.871388888888889,
      "maskedMaeBefore": 80.26666666666667,
      "maskedMaeImprovement": 0.9766853543743078,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 105.26112500519957,
      "inferenceMs": 94.5123749988852,
      "inputMegapixels": 0.049987,
      "megapixelsPerSecond": 0.5288937030795132,
      "maskedPixels": 1200,
      "paddedWidth": 264,
      "paddedHeight": 200,
      "driverAllocatedMemoryBytes": 1282228224,
      "totalMs": 105.26112500519957,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 0,
      "maskedMaeBefore": 0,
      "maskedMaeImprovement": 1,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": true,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 27.18954200099688,
      "inferenceMs": 23.28716700139921,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.8142538762255733,
      "maskedPixels": 0,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1282228224,
      "totalMs": 27.18954200099688,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 18.8625,
      "maskedMaeBefore": 104.15104166666667,
      "maskedMaeImprovement": 0.8188928339250887,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 126.85570798930712,
      "inferenceMs": 107.08320800040383,
      "inputMegapixels": 0.121728,
      "megapixelsPerSecond": 1.1367608635664048,
      "maskedPixels": 320,
      "paddedWidth": 384,
      "paddedHeight": 320,
      "driverAllocatedMemoryBytes": 1324105728,
      "totalMs": 126.85570798930712,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "maskedMae": 2.2216666666666667,
      "maskedMaeBefore": 79.71944444444445,
      "maskedMaeImprovement": 0.9721314331509808,
      "outsideMaskExact": true,
      "dimensionsExact": true,
      "emptyMaskIdentity": null,
      "gates": {
        "maskedMaeMaximum": 35,
        "maskedMaeImprovementMinimum": 0.6,
        "outsideMaskExact": true,
        "dimensionsExact": true,
        "emptyMaskIdentity": true
      }
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.10.0",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Entire LaMa graph including FFT on selected device; CPU decode, padding, compositing and PNG encoding",
      "durationMs": 26.127624994842336,
      "inferenceMs": 22.50274999823887,
      "inputMegapixels": 0.065536,
      "megapixelsPerSecond": 2.9123551568199013,
      "maskedPixels": 1200,
      "paddedWidth": 256,
      "paddedHeight": 256,
      "driverAllocatedMemoryBytes": 1324105728,
      "totalMs": 26.127624994842336,
      "requestedDevice": "mps",
      "precision": "fp32",
      "fallback": false,
      "grade": "fast",
      "artifactRevision": "b2456284b604786a88a38549432ddf2e73a5911f9868ea692b5122e89a5f1f54",
      "cpuThreads": 4
    }
  }
]

Something not quite right?

Tell us what happened with LaMa. Loading, results, or the page itself—we’re here for all of it.

Attachments Optional

Up to 3 files · 50 MB each

Browser, OS, CPU/GPU, memory, screen, storage, and network details are included automatically. Model inputs, results, and browsing history are not collected.