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.

NAFNet

Reduce image noise locally with the original NAFNet SIDD Width 32 checkpoint.

Desktop inferencePrivate · Runs on your computer
About this model

Reduce image noise locally with the original NAFNet SIDD Width 32 checkpoint.

Created by
Megvii
License
MIT
Model released
Apr 16, 2022
Source repository created
Apr 10, 2022
Model weights · FP32
116.9 MB

Best for

  • Denoise small RGB images locally
  • Review before/after restoration previews

Keep in mind

  • May smooth details, shift colors or add artifacts; review output before use. Not evidence-preserving restoration.
  • This checkpoint denoises; it does not deblur or upscale. Other checkpoints remain separate unavailable candidates.
  • PNG/JPEG still images only,32–512 pixels per edge, maximum8MB. No automatic resize or tiling. Output is opaque RGB PNG; EXIF orientation is applied and transparent inputs are composited onto white.
  • Quality measured on one photograph with two synthetic noise levels and a clean negative; real-camera noise, faces, text and larger images have no broad quality certification.
  • Single complete network output; runtime has no incremental image output and generateStream is not advertised.

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/megvii-research/NAFNet
Release verification
Verified
Release evidence
First official README revision publishing this exact SIDD Width32 checkpoint link: April16,2022. User explicitly requested this older model on2026-11-04; exception to2024+ selection policy. Source repository creation/update are separate. Original Google Drive Last-Modified:2022-04-15T07:11:13Z; first README publication is2022-04-16.
Source updated
Mar 28, 2024
Native model files
https://drive.usercontent.google.com/download?id=1lsByk21Xw-6aW7epCwOQxvm6HYCQZPHZ&export=download&confirm=t
Files updated
Apr 15, 2022
Artifact revision
89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4

Desktop app

Sample result

NAFNet Desktop

Run NAFNet 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.

NAFNet Desktop

21 configurations tested
Passed21
Failed0
TestedNov 4, 2026
Runtimepytorch 2.6.0

Quality criteria

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

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6456 ms / 0 ms1.173 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6457 ms / 0 ms1.145 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6457 ms / 0 ms1.144 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6458 ms / 618 ms1.138 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6458 ms / 0 ms1.136 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6459 ms / 0 ms1.111 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6459 ms / 0 ms1.109 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6460 ms / 618 ms1.091 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6462 ms / 0 ms1.061 MP/s
FP32mpsAcceleratedPassedVery fastmacos · aarch6496 ms / 617 ms0.684 MP/s
FP32cpuPassedVery fastmacos · aarch6492 ms / 220 ms0.714 MP/s
FP32cpuPassedVery fastmacos · aarch6492 ms / 0 ms0.712 MP/s
FP32cpuPassedVery fastmacos · aarch6493 ms / 0 ms0.708 MP/s
FP32cpuPassedVery fastmacos · aarch6493 ms / 0 ms0.704 MP/s
FP32cpuPassedVery fastmacos · aarch6493 ms / 0 ms0.703 MP/s
FP32cpuPassedVery fastmacos · aarch6493 ms / 0 ms0.703 MP/s
FP32cpuPassedVery fastmacos · aarch6493 ms / 0 ms0.701 MP/s
FP32cpuPassedVery fastmacos · aarch6496 ms / 0 ms0.681 MP/s
FP32cpuPassedVery fastmacos · aarch6496 ms / 220 ms0.681 MP/s
FP32cpuPassedFastmacos · aarch64101 ms / 219 ms0.650 MP/s
FP32cpuPassedFastmacos · aarch64195 ms / 301 ms0.337 MP/s
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-04T01:06:32.927647+00:00
Environment
{ "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2", "runtime": "PyTorch2.6.0 / NumPy2.3.4 / Pillow12.0.0", "interpreter": "App-owned CPython 3.13.18 model-private managed environment" }
Runtime
pytorch 2.6.0
Artifact revision
89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4
Fixture
{
  "file": "model-samples/desktop/nafnet/noise15.png",
  "size": 165113,
  "sha1": "a8b79ed7b3862f221adeb8ffc1ba3e5c150e9dcd",
  "sha256": "65e542f26478764e5757e4f1e4ed5483ed4b276821e1cac7d50006cc245ff655",
  "url": "https://assets.localai.tools/samples/v1/files/65e542f26478764e5757e4f1e4ed5483ed4b276821e1cac7d50006cc245ff655/noise15.png"
}
Rerun when
Checkpoint/runtime revision; Image preprocessing, network or output changes; Fixture/reference/quality gates; Relevant OS/backend behavior
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975869283902455,
      "gainDb": 4.334525193348892,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 300.9114580054302,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 194.51808399753645,
      "durationMs": 199.9845000100322,
      "megapixelsPerSecond": 0.33691469015718867,
      "imagesPerSecond": 5.140910189166087,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 661585920,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 500.8959580154624,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975869283902455,
      "gainDb": 4.334525193348892,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 220.37583400378935,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 91.8003749975469,
      "durationMs": 96.48808400379494,
      "megapixelsPerSecond": 0.7138968659087859,
      "imagesPerSecond": 10.893201689282012,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 643416064,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 316.8639180075843,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 20.33805748638473,
      "outputPsnrDb": 26.66768564821583,
      "gainDb": 6.329628161831099,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 92.52941701561213,
      "durationMs": 97.12241700617597,
      "megapixelsPerSecond": 0.7082720513514352,
      "imagesPerSecond": 10.807373830435719,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 643547136,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 97.12241700617597,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 120,
      "outputPsnrDb": 32.54052746815541,
      "gainDb": -87.4594725318446,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 96.26379201654345,
      "durationMs": 101.08650001347996,
      "megapixelsPerSecond": 0.6807959527372169,
      "imagesPerSecond": 10.388121837420913,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 644562944,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 101.08650001347996,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975869283902455,
      "gainDb": 4.334525193348892,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 93.16516600665636,
      "durationMs": 97.67283298424445,
      "megapixelsPerSecond": 0.7034388796700867,
      "imagesPerSecond": 10.733625483247174,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 644595712,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 97.67283298424445,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975869283902455,
      "gainDb": 4.334525193348892,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 93.17191698937677,
      "durationMs": 97.51479199621826,
      "megapixelsPerSecond": 0.703387910409445,
      "imagesPerSecond": 10.73284775405037,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 645939200,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 97.51479199621826,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975869283902455,
      "gainDb": 4.334525193348892,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 93.44641599454917,
      "durationMs": 97.83150002476759,
      "megapixelsPerSecond": 0.7013217072319048,
      "imagesPerSecond": 10.701319995604017,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 645955584,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 97.83150002476759,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975869283902455,
      "gainDb": 4.334525193348892,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 93.04887498728931,
      "durationMs": 97.29445798438974,
      "megapixelsPerSecond": 0.7043180265097495,
      "imagesPerSecond": 10.74704019942855,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 645971968,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 97.29445798438974,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975869283902455,
      "gainDb": 4.334525193348892,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 92.06379199167714,
      "durationMs": 96.53479198459536,
      "megapixelsPerSecond": 0.7118542326164956,
      "imagesPerSecond": 10.862033578742913,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 645971968,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 96.53479198459536,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975869283902455,
      "gainDb": 4.334525193348892,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 219.25295802066103,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 100.78879099455662,
      "durationMs": 105.57499999413267,
      "megapixelsPerSecond": 0.6502310361430911,
      "imagesPerSecond": 9.92173822239824,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 639041536,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 324.8279580147937,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975869283902455,
      "gainDb": 4.334525193348892,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 220.3664580010809,
      "placement": {
        "cpu": 664
      },
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 96.27383400220424,
      "durationMs": 101.08712498913519,
      "megapixelsPerSecond": 0.6807249413012836,
      "imagesPerSecond": 10.38703828889898,
      "visualizationMs": 0,
      "inputDevice": "cpu",
      "outputDevice": "cpu",
      "peakHelperRssBytes": 642220032,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 321.4535829902161,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975872235672533,
      "gainDb": 4.33452814511897,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 617.3450000060257,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 95.84287501638755,
      "durationMs": 127.74870899738744,
      "megapixelsPerSecond": 0.683785831641574,
      "imagesPerSecond": 10.433743768944916,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 543293440,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271037952,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 745.0937090034131,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 20.33805748638473,
      "outputPsnrDb": 26.667675396792355,
      "gainDb": 6.329617910407624,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 57.69195800530724,
      "durationMs": 64.36608301009983,
      "megapixelsPerSecond": 1.1359642186866177,
      "imagesPerSecond": 17.33343839548672,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 543703040,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271037952,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 64.36608301009983,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 120,
      "outputPsnrDb": 32.54054088279051,
      "gainDb": -87.45945911720949,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 57.24291698425077,
      "durationMs": 64.51216599089094,
      "megapixelsPerSecond": 1.1448752693373558,
      "imagesPerSecond": 17.469410237691587,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 543801344,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271037952,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 64.51216599089094,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975872235672533,
      "gainDb": 4.33452814511897,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 55.84833401371725,
      "durationMs": 62.228042021160945,
      "megapixelsPerSecond": 1.1734638312380687,
      "imagesPerSecond": 17.90563707333479,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 543883264,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271021568,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.228042021160945,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975872235672533,
      "gainDb": 4.33452814511897,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 61.74945901148021,
      "durationMs": 68.1669169862289,
      "megapixelsPerSecond": 1.0613210390687926,
      "imagesPerSecond": 16.19447386274403,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 543981568,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271021568,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 68.1669169862289,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975872235672533,
      "gainDb": 4.33452814511897,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 57.27912500151433,
      "durationMs": 63.27958399197087,
      "megapixelsPerSecond": 1.144151556055847,
      "imagesPerSecond": 17.458367249387315,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 544014336,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271021568,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 63.27958399197087,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975872235672533,
      "gainDb": 4.33452814511897,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 58.98324999725446,
      "durationMs": 65.08529098937288,
      "megapixelsPerSecond": 1.1110950990840713,
      "imagesPerSecond": 16.95396574530138,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 544129024,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271021568,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 65.08529098937288,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975872235672533,
      "gainDb": 4.33452814511897,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 0,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 59.073208016343415,
      "durationMs": 65.7695830159355,
      "megapixelsPerSecond": 1.109403098302509,
      "imagesPerSecond": 16.92814786228194,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 544178176,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271021568,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 65.7695830159355,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975872235672533,
      "gainDb": 4.33452814511897,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 617.9243329970632,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 60.0538750004489,
      "durationMs": 89.78425001259893,
      "megapixelsPerSecond": 1.0912867820687695,
      "imagesPerSecond": 16.65171481428176,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 542195712,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271037952,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 707.7085830096621,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "inputPsnrDb": 24.641344090553563,
      "outputPsnrDb": 28.975872235672533,
      "gainDb": 4.33452814511897,
      "passed": true
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "torchVersion": "2.6.0",
      "loadMs": 617.9470829956699,
      "placement": {
        "mps:0": 664
      },
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Network parameters, input and output tensors on selected device; PNG/JPEG decoding, normalization, PNG encoding and SVG presentation on CPU.",
      "inferenceMs": 57.612250006059185,
      "durationMs": 84.70191698870622,
      "megapixelsPerSecond": 1.1375358538003195,
      "imagesPerSecond": 17.357419644169916,
      "visualizationMs": 0,
      "inputDevice": "mps:0",
      "outputDevice": "mps:0",
      "peakHelperRssBytes": 540672000,
      "memoryScope": "Process lifetime peak macOS RSS, not GPU peak/minimum",
      "driverAllocatedMemoryBytes": 1271037952,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 702.6489999843761,
      "grade": "fast",
      "fallback": false,
      "loadTimingScope": "Model construction, checkpoint load and256px warmup after imports; excludes process/runtime/artifact download",
      "totalTimingScope": "Measured load plus complete image preprocessing/inference/encoding",
      "artifactRevision": "89c70e808d1783b6c07911306e106aaf0d4f7f3da8c61078b99ff7f8929a26f4"
    }
  }
]

Something not quite right?

Tell us what happened with NAFNet. 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.