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.

GLiNER 2.1

Find entities in text using user-supplied labels. Returns exact Unicode code-point spans, text, label and confidence through a local GLiNER/Transformers adapter.

Desktop inferencePrivate · Runs on your computer
About this model

Find entities in text using user-supplied labels. Returns exact Unicode code-point spans, text, label and confidence through a local GLiNER/Transformers adapter.

Created by
Urchade
License
Apache-2.0
Model released
Unknown
Source repository created
Apr 9, 2024
Model weights · FP32
1.2 GB

Best for

  • Interactive entity extraction with custom labels
  • Structuring notes and documents with human review
  • Multilingual entity annotation experiments

Keep in mind

  • Can miss entities, invent matches or assign the wrong type; scores are not calibrated correctness probabilities.
  • Multilingual training does not guarantee equal language/domain accuracy. Only a small authored English/French/Spanish suite is measured.
  • No privacy-redaction completeness, medical/legal/forensic accuracy or safety-critical classification guarantee.
  • The encoder uses native Transformers DeBERTa plus the installed GLiNER span head; this is not a plain fixed-label AutoModelForTokenClassification.
  • Text is processed in overlapping128-word windows with12-word overlap, adaptively shortened for encoder context. Boundary entities may be missed; maximum learned entity span is12words. No product text-length ceiling, but time/memory grow with input.
  • Very long labels or a single word that cannot fit the encoder context return validation errors rather than silent truncation.
  • Offsets count Unicode code points, end-exclusive; JavaScript highlighting must convert with Array.from rather than UTF16 slicing.
  • Only measured FP32 CPU/Metal pairs are advertised. Other precisions, Windows CUDA/AMD/Intel and other Mac versions/chips remain unverified or unsupported.
  • 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://huggingface.co/urchade/gliner_multi-v2.1
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
Dec 8, 2025
Native model files
https://huggingface.co/urchade/gliner_multi-v2.1
Files updated
Dec 8, 2025
Artifact revision
443d26d654e0324125a96bebd8e796c14ff2efe6

Desktop app

Sample result

GLiNER Multilingual v2.1 — Entity Extraction

Run GLiNER 2.1 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
macOS Apple SiliconMPS / FP16Unverified8.0 GiBRAM estimate
macOS Apple SiliconMPS / BF16Unverified8.0 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.

GLiNER Multilingual v2.1 — Entity Extraction

58 configurations tested
Passed58
Failed0
TestedNov 1, 2026
Runtimegliner 0.2.30

Quality criteria

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

PrecisionDeviceResultPlatformInference / load
FP16mpsAcceleratedPassedVery fastmacos · aarch6430 ms / 0 ms
FP16mpsAcceleratedPassedVery fastmacos · aarch6430 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6433 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6433 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms
BF16mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6439 ms / 0 ms
BF16mpsAcceleratedPassedVery fastmacos · aarch6440 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6444 ms / 0 ms
BF16mpsAcceleratedPassedVery fastmacos · aarch6445 ms / 0 ms
FP16mpsAcceleratedPassedVery fastmacos · aarch6446 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6458 ms / 0 ms
FP16mpsAcceleratedPassedVery fastmacos · aarch6483 ms / 0 ms
BF16mpsAcceleratedPassedVery fastmacos · aarch6496 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64110 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64119 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64258 ms / 6270 ms
FP32mpsAcceleratedPassedFastmacos · aarch64258 ms / 6296 ms
FP16mpsAcceleratedPassedFastmacos · aarch64264 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64304 ms / 0 ms
BF16mpsAcceleratedPassedFastmacos · aarch64305 ms / 0 ms
FP16mpsAcceleratedPassedFastmacos · aarch64509 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64535 ms / 0 ms
BF16mpsAcceleratedPassedFastmacos · aarch64544 ms / 0 ms
FP16mpsAcceleratedPassedFastmacos · aarch64790 ms / 0 ms
BF16mpsAcceleratedPassedFastmacos · aarch64838 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64891 ms / 0 ms
FP32mpsAcceleratedPassedModeratemacos · aarch641088 ms / 6156 ms
FP16mpsAcceleratedPassedModeratemacos · aarch641193 ms / 6350 ms
BF16mpsAcceleratedPassedModeratemacos · aarch641445 ms / 6428 ms
FP32mpsAcceleratedPassedModeratemacos · aarch642003 ms / 6239 ms
FP32cpuPassedVery fastmacos · aarch6433 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6435 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6435 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6436 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6436 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6436 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6437 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6437 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6437 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6438 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6438 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6438 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6439 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6439 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6439 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6439 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6439 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6440 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6441 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6441 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6456 ms / 6151 ms
FP32cpuPassedVery fastmacos · aarch6459 ms / 6248 ms
FP32cpuPassedVery fastmacos · aarch6460 ms / 6218 ms
FP32cpuPassedVery fastmacos · aarch6461 ms / 6211 ms
FP32cpuPassedFastmacos · aarch64526 ms / 6050 ms
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-01T18:11:53.195660+00:00
Environment
{ "hardware": "Apple M3 Max", "os": "macOS-27.2-arm64-arm-64bit-Mach-O", "python": "3.13.1", "runtime": "GLiNER0.2.30/Transformers5.22.1/Torch2.10.0" }
Runtime
gliner 0.2.30
Artifact revision
443d26d654e0324125a96bebd8e796c14ff2efe6
Fixture
{
  "file": "model-samples/desktop/gliner-english.txt",
  "size": 44,
  "sha1": "856394196a8f85ad18fca0d7812e4ab016b430b7",
  "sha256": "f5cda0441185c2a7b9e209c846c036420d4cfa1de2c5c1de20f801f0cb12c17c",
  "url": "https://assets.localai.tools/samples/v1/files/f5cda0441185c2a7b9e209c846c036420d4cfa1de2c5c1de20f801f0cb12c17c/gliner-english.txt"
}
Rerun when
artifact/encoder revisions/checksums; GLiNER/Transformers/Torch or driver/backend changes; window/tokenization/decoding/offset/config changes; fixtures/annotations/quality gates
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6248.320709000836,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "maxWordsPerChunk": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 59.383708001405466,
      "durationMs": 59.383708001405466,
      "charactersPerSecond": 740.943963939716,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 6307.704417002242,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Marie Curie",
          "label": "person",
          "start": 0,
          "end": 11
        },
        {
          "text": "Varsovie",
          "label": "location",
          "start": 22,
          "end": 30
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 48,
          "end": 53
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "maxWordsPerChunk": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 37.68712500095717,
      "durationMs": 37.68712500095717,
      "charactersPerSecond": 1432.8500780738389,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 54,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 37.68712500095717,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "María García",
          "label": "person",
          "start": 0,
          "end": 12
        },
        {
          "text": "Google",
          "label": "organization",
          "start": 24,
          "end": 30
        },
        {
          "text": "Madrid",
          "label": "location",
          "start": 34,
          "end": 40
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "maxWordsPerChunk": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 36.06720799871255,
      "durationMs": 36.06720799871255,
      "charactersPerSecond": 1136.766671860587,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 41,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 36.06720799871255,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "maxWordsPerChunk": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 34.94795900041936,
      "durationMs": 34.94795900041936,
      "charactersPerSecond": 858.419228420178,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 34.94795900041936,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [],
      "validOffsets": true,
      "exactNegative": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "maxWordsPerChunk": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 35.73479199985741,
      "durationMs": 35.73479199985741,
      "charactersPerSecond": 1231.293021103231,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 35.73479199985741,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "maxWordsPerChunk": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 35.82641699904343,
      "durationMs": 35.82641699904343,
      "charactersPerSecond": 1228.1440257108268,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 35.82641699904343,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "maxWordsPerChunk": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 36.68758300045738,
      "durationMs": 36.68758300045738,
      "charactersPerSecond": 1199.3158557065876,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 36.68758300045738,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "maxWordsPerChunk": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.603542001510505,
      "durationMs": 38.603542001510505,
      "charactersPerSecond": 1139.791783828498,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 38.603542001510505,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6218.163707999338,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 60.1160419973894,
      "durationMs": 60.1160419973894,
      "charactersPerSecond": 731.9177799814356,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 6278.279749996727,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Marie Curie",
          "label": "person",
          "start": 0,
          "end": 11
        },
        {
          "text": "Varsovie",
          "label": "location",
          "start": 22,
          "end": 30
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 48,
          "end": 53
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 41.469457999482984,
      "durationMs": 41.469457999482984,
      "charactersPerSecond": 1302.1631486158617,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 54,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 41.469457999482984,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "María García",
          "label": "person",
          "start": 0,
          "end": 12
        },
        {
          "text": "Google",
          "label": "organization",
          "start": 24,
          "end": 30
        },
        {
          "text": "Madrid",
          "label": "location",
          "start": 34,
          "end": 40
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 39.369958998577204,
      "durationMs": 39.369958998577204,
      "charactersPerSecond": 1041.4031673612285,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 41,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 39.369958998577204,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.28041700035101,
      "durationMs": 38.28041700035101,
      "charactersPerSecond": 783.6905224863386,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.28041700035101,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [],
      "validOffsets": true,
      "exactNegative": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 40.70824999871547,
      "durationMs": 40.70824999871547,
      "charactersPerSecond": 1080.8619874690853,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 40.70824999871547,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.92904200256453,
      "durationMs": 38.92904200256453,
      "charactersPerSecond": 1130.261566598567,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.92904200256453,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 39.23970899995766,
      "durationMs": 39.23970899995766,
      "charactersPerSecond": 1121.3131065790392,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 39.23970899995766,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 40.4486250008631,
      "durationMs": 40.4486250008631,
      "charactersPerSecond": 1087.799647059971,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 40.4486250008631,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6211.163333999139,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 61.20287500016275,
      "durationMs": 61.20287500016275,
      "charactersPerSecond": 718.9204755476437,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 6272.366208999301,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.25029200015706,
      "durationMs": 38.25029200015706,
      "charactersPerSecond": 1150.3180158682012,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.25029200015706,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 37.405833001685096,
      "durationMs": 37.405833001685096,
      "charactersPerSecond": 1176.2871314219321,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.405833001685096,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6151.181790999544,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 56.16662499960512,
      "durationMs": 56.16662499960512,
      "charactersPerSecond": 783.3833704679486,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 6207.348415999149,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.744834000681294,
      "durationMs": 38.744834000681294,
      "charactersPerSecond": 1135.6352694458906,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.744834000681294,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 36.945291998563334,
      "durationMs": 36.945291998563334,
      "charactersPerSecond": 1190.9501216477324,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.945291998563334,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6238.996749998478,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 2003.3126670023194,
      "durationMs": 2003.3126670023194,
      "charactersPerSecond": 21.963620918865313,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1200816128,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 8242.309417000797,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "moderate"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Marie Curie",
          "label": "person",
          "start": 0,
          "end": 11
        },
        {
          "text": "Varsovie",
          "label": "location",
          "start": 22,
          "end": 30
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 48,
          "end": 53
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 890.6278329995985,
      "durationMs": 890.6278329995985,
      "charactersPerSecond": 60.631386084275164,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 54,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1209794560,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 890.6278329995985,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "María García",
          "label": "person",
          "start": 0,
          "end": 12
        },
        {
          "text": "Google",
          "label": "organization",
          "start": 24,
          "end": 30
        },
        {
          "text": "Madrid",
          "label": "location",
          "start": 34,
          "end": 40
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 110.18616699948325,
      "durationMs": 110.18616699948325,
      "charactersPerSecond": 372.0975247300533,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 41,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1209909248,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 110.18616699948325,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 534.8330830020132,
      "durationMs": 534.8330830020132,
      "charactersPerSecond": 56.092266827643265,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1210269696,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 534.8330830020132,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [],
      "validOffsets": true,
      "exactNegative": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 304.28037499950733,
      "durationMs": 304.28037499950733,
      "charactersPerSecond": 144.60347631710144,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1210662912,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 304.28037499950733,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 57.63262500113342,
      "durationMs": 57.63262500113342,
      "charactersPerSecond": 763.4564623619813,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1210662912,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 57.63262500113342,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 35.8825839975907,
      "durationMs": 35.8825839975907,
      "charactersPerSecond": 1226.2216122159523,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1210662912,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.8825839975907,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 36.92129100090824,
      "durationMs": 36.92129100090824,
      "charactersPerSecond": 1191.7243088525163,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1210662912,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.92129100090824,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6269.709042000613,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 257.6174579990038,
      "durationMs": 257.6174579990038,
      "charactersPerSecond": 170.795878283102,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1200832512,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 6527.326499999617,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.51687499991385,
      "durationMs": 38.51687499991385,
      "charactersPerSecond": 1142.3564346821597,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1200832512,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.51687499991385,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 32.51941599955899,
      "durationMs": 32.51941599955899,
      "charactersPerSecond": 1353.037828249951,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1200832512,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 32.51941599955899,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6296.28537499957,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 257.82558399805566,
      "durationMs": 257.82558399805566,
      "charactersPerSecond": 170.6580057638183,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1200832512,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 6554.110958997626,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 35.80425000109244,
      "durationMs": 35.80425000109244,
      "charactersPerSecond": 1228.9043898044922,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1200832512,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.80425000109244,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 33.458416000939906,
      "durationMs": 33.458416000939906,
      "charactersPerSecond": 1315.0652439363525,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1200832512,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 33.458416000939906,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6350.316000000021,
      "placement": {
        "mps:0/torch.float16": 224
      },
      "parameterDtype": "torch.float16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 1193.3229999995092,
      "durationMs": 1193.3229999995092,
      "charactersPerSecond": 36.87182766109268,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1119027200,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp16",
      "cpuThreads": 4,
      "totalMs": 7543.63899999953,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "moderate"
    }
  },
  {
    "precision": "fp16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Marie Curie",
          "label": "person",
          "start": 0,
          "end": 11
        },
        {
          "text": "Varsovie",
          "label": "location",
          "start": 22,
          "end": 30
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 48,
          "end": 53
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float16": 224
      },
      "parameterDtype": "torch.float16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 790.45487500116,
      "durationMs": 790.45487500116,
      "charactersPerSecond": 68.31509515318095,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 54,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1119617024,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp16",
      "cpuThreads": 4,
      "totalMs": 790.45487500116,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "María García",
          "label": "person",
          "start": 0,
          "end": 12
        },
        {
          "text": "Google",
          "label": "organization",
          "start": 24,
          "end": 30
        },
        {
          "text": "Madrid",
          "label": "location",
          "start": 34,
          "end": 40
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float16": 224
      },
      "parameterDtype": "torch.float16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 82.75291700192611,
      "durationMs": 82.75291700192611,
      "charactersPerSecond": 495.4508129187242,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 41,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1119731712,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp16",
      "cpuThreads": 4,
      "totalMs": 82.75291700192611,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float16": 224
      },
      "parameterDtype": "torch.float16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 509.2052500003774,
      "durationMs": 509.2052500003774,
      "charactersPerSecond": 58.915339148560946,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120092160,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp16",
      "cpuThreads": 4,
      "totalMs": 509.2052500003774,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [],
      "validOffsets": true,
      "exactNegative": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float16": 224
      },
      "parameterDtype": "torch.float16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 264.0220829998725,
      "durationMs": 264.0220829998725,
      "charactersPerSecond": 166.65272654492787,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120485376,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp16",
      "cpuThreads": 4,
      "totalMs": 264.0220829998725,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float16": 224
      },
      "parameterDtype": "torch.float16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 46.28304099969682,
      "durationMs": 46.28304099969682,
      "charactersPerSecond": 950.6721911442297,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120485376,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp16",
      "cpuThreads": 4,
      "totalMs": 46.28304099969682,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float16": 224
      },
      "parameterDtype": "torch.float16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 30.019083998922724,
      "durationMs": 30.019083998922724,
      "charactersPerSecond": 1465.7342642959725,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120485376,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp16",
      "cpuThreads": 4,
      "totalMs": 30.019083998922724,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float16": 224
      },
      "parameterDtype": "torch.float16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 29.6615830011433,
      "durationMs": 29.6615830011433,
      "charactersPerSecond": 1483.40026216079,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120485376,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp16",
      "cpuThreads": 4,
      "totalMs": 29.6615830011433,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "bf16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6428.343916999438,
      "placement": {
        "mps:0/torch.bfloat16": 224
      },
      "parameterDtype": "torch.bfloat16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 1444.5936669981165,
      "durationMs": 1444.5936669981165,
      "charactersPerSecond": 30.458391868373994,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1119027200,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "bf16",
      "cpuThreads": 4,
      "totalMs": 7872.937583997555,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "moderate"
    }
  },
  {
    "precision": "bf16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Marie Curie",
          "label": "person",
          "start": 0,
          "end": 11
        },
        {
          "text": "Varsovie",
          "label": "location",
          "start": 22,
          "end": 30
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 48,
          "end": 53
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.bfloat16": 224
      },
      "parameterDtype": "torch.bfloat16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 838.2989160018042,
      "durationMs": 838.2989160018042,
      "charactersPerSecond": 64.41616345819513,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 54,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1119617024,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "bf16",
      "cpuThreads": 4,
      "totalMs": 838.2989160018042,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "bf16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "María García",
          "label": "person",
          "start": 0,
          "end": 12
        },
        {
          "text": "Google",
          "label": "organization",
          "start": 24,
          "end": 30
        },
        {
          "text": "Madrid",
          "label": "location",
          "start": 34,
          "end": 40
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.bfloat16": 224
      },
      "parameterDtype": "torch.bfloat16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 95.62408400233835,
      "durationMs": 95.62408400233835,
      "charactersPerSecond": 428.7622770744387,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 41,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1119731712,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "bf16",
      "cpuThreads": 4,
      "totalMs": 95.62408400233835,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "bf16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.bfloat16": 224
      },
      "parameterDtype": "torch.bfloat16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 544.2647500021849,
      "durationMs": 544.2647500021849,
      "charactersPerSecond": 55.1202333053529,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120092160,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "bf16",
      "cpuThreads": 4,
      "totalMs": 544.2647500021849,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "bf16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [],
      "validOffsets": true,
      "exactNegative": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.bfloat16": 224
      },
      "parameterDtype": "torch.bfloat16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 305.3010000003269,
      "durationMs": 305.3010000003269,
      "charactersPerSecond": 144.12006511591147,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120485376,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "bf16",
      "cpuThreads": 4,
      "totalMs": 305.3010000003269,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "bf16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.bfloat16": 224
      },
      "parameterDtype": "torch.bfloat16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 45.37595900183078,
      "durationMs": 45.37595900183078,
      "charactersPerSecond": 969.676475558891,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120485376,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "bf16",
      "cpuThreads": 4,
      "totalMs": 45.37595900183078,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "bf16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.bfloat16": 224
      },
      "parameterDtype": "torch.bfloat16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 40.46216600181651,
      "durationMs": 40.46216600181651,
      "charactersPerSecond": 1087.4356058453386,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120485376,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "bf16",
      "cpuThreads": 4,
      "totalMs": 40.46216600181651,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "bf16",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true,
      "exactNegative": null
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.bfloat16": 224
      },
      "parameterDtype": "torch.bfloat16",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 36.399500000698026,
      "durationMs": 36.399500000698026,
      "charactersPerSecond": 1208.8078132709575,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1120485376,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "bf16",
      "cpuThreads": 4,
      "totalMs": 36.399500000698026,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ],
      "validOffsets": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6050.18820800251,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 525.814790998993,
      "durationMs": 525.814790998993,
      "charactersPerSecond": 7219.272004098626,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 6576.002999001503,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [],
      "validOffsets": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 35.07066599922837,
      "durationMs": 35.07066599922837,
      "charactersPerSecond": 1254.6097642105826,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 1,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.07066599922837,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 33.41350000118837,
      "durationMs": 33.41350000118837,
      "charactersPerSecond": 1316.833016548255,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": false,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 33.41350000118837,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ],
      "validOffsets": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6155.970707997767,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 1087.564208999538,
      "durationMs": 1087.564208999538,
      "charactersPerSecond": 3490.368631652545,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2352021504,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 7243.534916997305,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "moderate"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [],
      "validOffsets": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 119.22295900149038,
      "durationMs": 119.22295900149038,
      "charactersPerSecond": 369.0564331610824,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 1,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2352349184,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 119.22295900149038,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ],
      "validOffsets": true
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 43.57929099933244,
      "durationMs": 43.57929099933244,
      "charactersPerSecond": 1009.6538743751935,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": false,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2352398336,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 43.57929099933244,
      "artifactRevision": "443d26d654e0324125a96bebd8e796c14ff2efe6",
      "fallback": false,
      "grade": "fast"
    }
  }
]

GLiNER Large English v2.1

36 configurations tested
Passed36
Failed0
TestedNov 6, 2026
Runtimegliner 0.2.30

Quality criteria

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

PrecisionDeviceResultPlatformInference / load
FP32mpsAcceleratedPassedVery fastmacos · aarch6449 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6450 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6452 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6453 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6453 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6458 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64114 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64259 ms / 7252 ms
FP32mpsAcceleratedPassedFastmacos · aarch64270 ms / 7253 ms
FP32mpsAcceleratedPassedFastmacos · aarch64344 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64370 ms / 7350 ms
FP32mpsAcceleratedPassedFastmacos · aarch64760 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6497 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6498 ms / 0 ms
FP32cpuPassedFastmacos · aarch64102 ms / 0 ms
FP32cpuPassedFastmacos · aarch64103 ms / 0 ms
FP32cpuPassedFastmacos · aarch64104 ms / 0 ms
FP32cpuPassedFastmacos · aarch64105 ms / 0 ms
FP32cpuPassedFastmacos · aarch64108 ms / 0 ms
FP32cpuPassedFastmacos · aarch64110 ms / 0 ms
FP32cpuPassedFastmacos · aarch64112 ms / 0 ms
FP32cpuPassedFastmacos · aarch64112 ms / 0 ms
FP32cpuPassedFastmacos · aarch64112 ms / 0 ms
FP32cpuPassedFastmacos · aarch64112 ms / 0 ms
FP32cpuPassedFastmacos · aarch64114 ms / 0 ms
FP32cpuPassedFastmacos · aarch64115 ms / 0 ms
FP32cpuPassedFastmacos · aarch64116 ms / 0 ms
FP32cpuPassedFastmacos · aarch64117 ms / 0 ms
FP32cpuPassedFastmacos · aarch64118 ms / 7105 ms
FP32cpuPassedFastmacos · aarch64126 ms / 6997 ms
FP32cpuPassedFastmacos · aarch64127 ms / 7057 ms
FP32cpuPassedFastmacos · aarch64132 ms / 6990 ms
FP32cpuPassedFastmacos · aarch64132 ms / 6998 ms
FP32cpuPassedFastmacos · aarch64133 ms / 7039 ms
FP32cpuPassedModeratemacos · aarch641238 ms / 0 ms
FP32cpuPassedModeratemacos · aarch641271 ms / 0 ms
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-06T12:18:32.736771+00:00
Environment
{ "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2", "python": "App-owned CPython3.13.18 private model probe venv" }
Runtime
gliner 0.2.30
Artifact revision
abd49a1f1ebc12af1be84d06f6848221cf96dcad
Fixture
{
  "file": "model-samples/desktop/gliner-english.txt",
  "size": 44,
  "sha1": "856394196a8f85ad18fca0d7812e4ab016b430b7",
  "sha256": "f5cda0441185c2a7b9e209c846c036420d4cfa1de2c5c1de20f801f0cb12c17c",
  "url": "https://assets.localai.tools/samples/v1/files/f5cda0441185c2a7b9e209c846c036420d4cfa1de2c5c1de20f801f0cb12c17c/gliner-english.txt"
}
Rerun when
artifact/encoder revisions/checksums; GLiNER/Transformers/Torch or driver/backend changes; window/tokenization/decoding/offset/config changes; fixtures/annotations/quality gates
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 7057.109583000056,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 126.82395799947699,
      "durationMs": 126.82395799947699,
      "charactersPerSecond": 346.93760306851055,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 7183.933540999533,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 105.49587500008784,
      "durationMs": 105.49587500008784,
      "charactersPerSecond": 284.37130835660656,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 105.49587500008784,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": true,
      "expected": []
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 101.79579199939326,
      "durationMs": 101.79579199939326,
      "charactersPerSecond": 432.23790626101965,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 101.79579199939326,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 1271.4890830002332,
      "durationMs": 1271.4890830002332,
      "charactersPerSecond": 2985.4758886665986,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 1271.4890830002332,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "moderate"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 104.06204199989588,
      "durationMs": 104.06204199989588,
      "charactersPerSecond": 422.8246837597519,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 104.06204199989588,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 102.56074999961129,
      "durationMs": 102.56074999961129,
      "charactersPerSecond": 429.01402339751576,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 102.56074999961129,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 7104.605792000257,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 117.67579199931788,
      "durationMs": 117.67579199931788,
      "charactersPerSecond": 373.90867953754713,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 7222.281583999575,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 98.37808300017059,
      "durationMs": 98.37808300017059,
      "charactersPerSecond": 447.2540901200901,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 98.37808300017059,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 96.82304200032377,
      "durationMs": 96.82304200032377,
      "charactersPerSecond": 454.43728157036077,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 96.82304200032377,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6989.801624999927,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 131.5348329999324,
      "durationMs": 131.5348329999324,
      "charactersPerSecond": 334.5121516216363,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 7121.33645799986,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 112.49733299973741,
      "durationMs": 112.49733299973741,
      "charactersPerSecond": 391.1203832725768,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 112.49733299973741,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 111.81725000005827,
      "durationMs": 111.81725000005827,
      "charactersPerSecond": 393.4992141192622,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 111.81725000005827,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 7039.037792000272,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 132.66379200013034,
      "durationMs": 132.66379200013034,
      "charactersPerSecond": 331.66547809787295,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 7171.701584000402,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 116.51562499991996,
      "durationMs": 116.51562499991996,
      "charactersPerSecond": 257.4761968621857,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 116.51562499991996,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": true,
      "expected": []
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 115.88808299984521,
      "durationMs": 115.88808299984521,
      "charactersPerSecond": 379.67665752188486,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 115.88808299984521,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 1237.8327919996082,
      "durationMs": 1237.8327919996082,
      "charactersPerSecond": 3066.6500552695015,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 1237.8327919996082,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "moderate"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 111.57408300005045,
      "durationMs": 111.57408300005045,
      "charactersPerSecond": 394.3568149243055,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 111.57408300005045,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 113.6228340001253,
      "durationMs": 113.6228340001253,
      "charactersPerSecond": 387.24610583073013,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 113.6228340001253,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6996.746417000395,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 126.2561249995997,
      "durationMs": 126.2561249995997,
      "charactersPerSecond": 348.4979441602497,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 7123.002541999995,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 109.5603330004451,
      "durationMs": 109.5603330004451,
      "charactersPerSecond": 401.60520505009094,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 109.5603330004451,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 107.79066600025544,
      "durationMs": 107.79066600025544,
      "charactersPerSecond": 408.1986097005443,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 107.79066600025544,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 6997.866624999915,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 131.9112090004637,
      "durationMs": 131.9112090004637,
      "charactersPerSecond": 333.55770395406904,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 7129.777834000379,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 114.74095799985662,
      "durationMs": 114.74095799985662,
      "charactersPerSecond": 383.47248242475877,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 114.74095799985662,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 111.84479199982889,
      "durationMs": 111.84479199982889,
      "charactersPerSecond": 393.40231416468026,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 111.84479199982889,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 7349.696792000032,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 370.3665839993846,
      "durationMs": 370.3665839993846,
      "charactersPerSecond": 118.80121452877377,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2184413184,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 7720.063375999416,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 344.1670829997747,
      "durationMs": 344.1670829997747,
      "charactersPerSecond": 87.1669647733849,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2184839168,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 344.1670829997747,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": true,
      "expected": []
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 114.41620799996599,
      "durationMs": 114.41620799996599,
      "charactersPerSecond": 384.5609006725086,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2202009600,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 114.41620799996599,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 760.0961250000182,
      "durationMs": 760.0961250000182,
      "charactersPerSecond": 4994.105186366934,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2209628160,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 760.0961250000182,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 57.650540999929945,
      "durationMs": 57.650540999929945,
      "charactersPerSecond": 763.2192037894921,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2209628160,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 57.650540999929945,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 50.22729200027243,
      "durationMs": 50.22729200027243,
      "charactersPerSecond": 876.0177634056271,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2209628160,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 50.22729200027243,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 7251.996957999836,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 259.2595420001089,
      "durationMs": 259.2595420001089,
      "charactersPerSecond": 169.71410062886525,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2184413184,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 7511.256499999945,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 53.26091699953395,
      "durationMs": 53.26091699953395,
      "charactersPerSecond": 826.1217132327072,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2184413184,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 53.26091699953395,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 51.7212919994563,
      "durationMs": 51.7212919994563,
      "charactersPerSecond": 850.7134740652367,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2184413184,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 51.7212919994563,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 7253.330874999847,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 269.71108300040214,
      "durationMs": 269.71108300040214,
      "charactersPerSecond": 163.13753039186156,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2184413184,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 7523.041958000249,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 52.714124999511114,
      "durationMs": 52.714124999511114,
      "charactersPerSecond": 834.6908916805139,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2184413184,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 52.714124999511114,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 416
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 48.8961669998389,
      "durationMs": 48.8961669998389,
      "charactersPerSecond": 899.8660365370761,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 2184413184,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 48.8961669998389,
      "artifactRevision": "abd49a1f1ebc12af1be84d06f6848221cf96dcad",
      "fallback": false,
      "grade": "fast"
    }
  }
]

GLiNER Medium English v2.1

36 configurations tested
Passed36
Failed0
TestedNov 1, 2026
Runtimegliner 0.2.30

Quality criteria

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

PrecisionDeviceResultPlatformInference / load
FP32mpsAcceleratedPassedVery fastmacos · aarch6429 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6430 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6431 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6438 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6491 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64130 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64214 ms / 3717 ms
FP32mpsAcceleratedPassedFastmacos · aarch64217 ms / 3686 ms
FP32mpsAcceleratedPassedFastmacos · aarch64224 ms / 3681 ms
FP32mpsAcceleratedPassedFastmacos · aarch64494 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6432 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6432 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6434 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6434 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6434 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6434 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6434 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6434 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6434 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6435 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6435 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6435 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6435 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6436 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6437 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6438 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6451 ms / 3640 ms
FP32cpuPassedVery fastmacos · aarch6452 ms / 3637 ms
FP32cpuPassedVery fastmacos · aarch6453 ms / 3615 ms
FP32cpuPassedVery fastmacos · aarch6453 ms / 3636 ms
FP32cpuPassedVery fastmacos · aarch6454 ms / 3576 ms
FP32cpuPassedVery fastmacos · aarch6457 ms / 3610 ms
FP32cpuPassedFastmacos · aarch64482 ms / 0 ms
FP32cpuPassedFastmacos · aarch64497 ms / 0 ms
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-01T22:50:16.468941+00:00
Environment
{ "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2", "python": "App-owned CPython3.13.18 private model probe venv" }
Runtime
gliner 0.2.30
Artifact revision
40ec419335d09393f298636f471328b722c6da9e
Fixture
{
  "file": "model-samples/desktop/gliner-english.txt",
  "size": 44,
  "sha1": "856394196a8f85ad18fca0d7812e4ab016b430b7",
  "sha256": "f5cda0441185c2a7b9e209c846c036420d4cfa1de2c5c1de20f801f0cb12c17c",
  "url": "https://assets.localai.tools/samples/v1/files/f5cda0441185c2a7b9e209c846c036420d4cfa1de2c5c1de20f801f0cb12c17c/gliner-english.txt"
}
Rerun when
artifact/encoder revisions/checksums; GLiNER/Transformers/Torch or driver/backend changes; window/tokenization/decoding/offset/config changes; fixtures/annotations/quality gates
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3639.8616250007763,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 51.49154199898476,
      "durationMs": 51.49154199898476,
      "charactersPerSecond": 854.5092706850289,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3691.353166999761,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 33.873832995595876,
      "durationMs": 33.873832995595876,
      "charactersPerSecond": 885.6393666432867,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 33.873832995595876,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": true,
      "expected": []
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 33.98270800244063,
      "durationMs": 33.98270800244063,
      "charactersPerSecond": 1294.7761548855945,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 33.98270800244063,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 497.25000000034925,
      "durationMs": 497.25000000034925,
      "charactersPerSecond": 7633.986928099213,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 497.25000000034925,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 32.434916000056546,
      "durationMs": 32.434916000056546,
      "charactersPerSecond": 1356.5627856080555,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 32.434916000056546,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 32.45062499627238,
      "durationMs": 32.45062499627238,
      "charactersPerSecond": 1355.9060882511292,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 32.45062499627238,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3614.5512080038316,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 52.59904199920129,
      "durationMs": 52.59904199920129,
      "charactersPerSecond": 836.5171365795624,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3667.150250003033,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 34.112458997697104,
      "durationMs": 34.112458997697104,
      "charactersPerSecond": 1289.8513121839264,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 34.112458997697104,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 34.11341700120829,
      "durationMs": 34.11341700120829,
      "charactersPerSecond": 1289.8150894248304,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 34.11341700120829,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3637.1815420061466,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 52.17566699866438,
      "durationMs": 52.17566699866438,
      "charactersPerSecond": 843.3049835496369,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3689.357209004811,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 34.72758400312159,
      "durationMs": 34.72758400312159,
      "charactersPerSecond": 1267.004350087957,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 34.72758400312159,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 33.83116600161884,
      "durationMs": 33.83116600161884,
      "charactersPerSecond": 1300.5759245157135,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 33.83116600161884,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3609.5106249995297,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 56.89745899871923,
      "durationMs": 56.89745899871923,
      "charactersPerSecond": 773.3210019271766,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3666.408083998249,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.33729100006167,
      "durationMs": 38.33729100006167,
      "charactersPerSecond": 782.5279047481927,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.33729100006167,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": true,
      "expected": []
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 35.28283299965551,
      "durationMs": 35.28283299965551,
      "charactersPerSecond": 1247.065392975377,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.28283299965551,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 482.2921250015497,
      "durationMs": 482.2921250015497,
      "charactersPerSecond": 7870.748459738592,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 482.2921250015497,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 35.61654099758016,
      "durationMs": 35.61654099758016,
      "charactersPerSecond": 1235.3810551953775,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.61654099758016,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 34.31662499497179,
      "durationMs": 34.31662499497179,
      "charactersPerSecond": 1282.177370485794,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 34.31662499497179,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3575.8800840048934,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 54.27941600646591,
      "durationMs": 54.27941600646591,
      "charactersPerSecond": 810.6203647209212,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3630.1595000113593,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 36.70737500215182,
      "durationMs": 36.70737500215182,
      "charactersPerSecond": 1198.6692046876324,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.70737500215182,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 34.82433399767615,
      "durationMs": 34.82433399767615,
      "charactersPerSecond": 1263.484321134071,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 34.82433399767615,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3635.607500000333,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 53.18841600092128,
      "durationMs": 53.18841600092128,
      "charactersPerSecond": 827.247797701625,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3688.795916001254,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 35.06487499544164,
      "durationMs": 35.06487499544164,
      "charactersPerSecond": 1254.8169644329234,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.06487499544164,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 33.964124995691236,
      "durationMs": 33.964124995691236,
      "charactersPerSecond": 1295.4845739609643,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 33.964124995691236,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3686.4546249998966,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 216.99579199776053,
      "durationMs": 216.99579199776053,
      "charactersPerSecond": 202.76890899549838,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3903.450416997657,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 129.98125000012806,
      "durationMs": 129.98125000012806,
      "charactersPerSecond": 230.80251959394485,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102708736,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 129.98125000012806,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": true,
      "expected": []
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 90.76154200010933,
      "durationMs": 90.76154200010933,
      "charactersPerSecond": 484.78682744225523,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1111490560,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 90.76154200010933,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 493.84187500254484,
      "durationMs": 493.84187500254484,
      "charactersPerSecond": 7686.670961186592,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1127088128,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 493.84187500254484,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 36.68933299923083,
      "durationMs": 36.68933299923083,
      "charactersPerSecond": 1199.258651034142,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1127088128,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.68933299923083,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 29.77258300234098,
      "durationMs": 29.77258300234098,
      "charactersPerSecond": 1477.8697567671686,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1127088128,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 29.77258300234098,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3717.1747499960475,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 214.18816700315801,
      "durationMs": 214.18816700315801,
      "charactersPerSecond": 205.42684787694765,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3931.3629169992055,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.015375001123175,
      "durationMs": 38.015375001123175,
      "charactersPerSecond": 1157.4264359801793,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.015375001123175,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 29.053124999336433,
      "durationMs": 29.053124999336433,
      "charactersPerSecond": 1514.4670324106253,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 29.053124999336433,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3681.2651660002302,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 223.9097089986899,
      "durationMs": 223.9097089986899,
      "charactersPerSecond": 196.507780733427,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3905.17487499892,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 35.598665999714285,
      "durationMs": 35.598665999714285,
      "charactersPerSecond": 1236.0013715219875,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.598665999714285,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 224
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 30.621707999671344,
      "durationMs": 30.621707999671344,
      "charactersPerSecond": 1436.8891506793886,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 30.621707999671344,
      "artifactRevision": "40ec419335d09393f298636f471328b722c6da9e",
      "fallback": false,
      "grade": "fast"
    }
  }
]

GLiNER Small English v2.1

36 configurations tested
Passed36
Failed0
TestedNov 1, 2026
Runtimegliner 0.2.30

Quality criteria

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

PrecisionDeviceResultPlatformInference / load
FP32mpsAcceleratedPassedVery fastmacos · aarch6424 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6425 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6427 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6429 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6433 ms / 0 ms
FP32mpsAcceleratedPassedVery fastmacos · aarch6441 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64213 ms / 3315 ms
FP32mpsAcceleratedPassedFastmacos · aarch64226 ms / 3271 ms
FP32mpsAcceleratedPassedFastmacos · aarch64227 ms / 0 ms
FP32mpsAcceleratedPassedFastmacos · aarch64307 ms / 3323 ms
FP32mpsAcceleratedPassedModeratemacos · aarch641188 ms / 0 ms
FP32mpsAcceleratedPassedModeratemacos · aarch641491 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6418 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6419 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6419 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6419 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6420 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6420 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6420 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6420 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6420 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6420 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6420 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6420 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6420 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6421 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6421 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6421 ms / 0 ms
FP32cpuPassedVery fastmacos · aarch6437 ms / 3248 ms
FP32cpuPassedVery fastmacos · aarch6438 ms / 3224 ms
FP32cpuPassedVery fastmacos · aarch6438 ms / 3183 ms
FP32cpuPassedVery fastmacos · aarch6438 ms / 3193 ms
FP32cpuPassedVery fastmacos · aarch6439 ms / 3237 ms
FP32cpuPassedVery fastmacos · aarch6442 ms / 4026 ms
FP32cpuPassedFastmacos · aarch64286 ms / 0 ms
FP32cpuPassedFastmacos · aarch64307 ms / 0 ms
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-11-01T22:49:04.174481+00:00
Environment
{ "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2", "python": "App-owned CPython3.13.18 private model probe venv" }
Runtime
gliner 0.2.30
Artifact revision
4e091416cf7c3481db542c2a3d26156916f3a47f
Fixture
{
  "file": "model-samples/desktop/gliner-english.txt",
  "size": 44,
  "sha1": "856394196a8f85ad18fca0d7812e4ab016b430b7",
  "sha256": "f5cda0441185c2a7b9e209c846c036420d4cfa1de2c5c1de20f801f0cb12c17c",
  "url": "https://assets.localai.tools/samples/v1/files/f5cda0441185c2a7b9e209c846c036420d4cfa1de2c5c1de20f801f0cb12c17c/gliner-english.txt"
}
Rerun when
artifact/encoder revisions/checksums; GLiNER/Transformers/Torch or driver/backend changes; window/tokenization/decoding/offset/config changes; fixtures/annotations/quality gates
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 4026.0593330021948,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 42.3563340009423,
      "durationMs": 42.3563340009423,
      "charactersPerSecond": 1038.8056718747457,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 4068.415667003137,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 20.926375000271946,
      "durationMs": 20.926375000271946,
      "charactersPerSecond": 1433.5975533082121,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 20.926375000271946,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": true,
      "expected": []
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 19.92979099304648,
      "durationMs": 19.92979099304648,
      "charactersPerSecond": 2207.750197448214,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 19.92979099304648,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 306.64129099750426,
      "durationMs": 306.64129099750426,
      "charactersPerSecond": 12379.285215150283,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 306.64129099750426,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 20.207375004247297,
      "durationMs": 20.207375004247297,
      "charactersPerSecond": 2177.4228463989925,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 20.207375004247297,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 19.553500002075452,
      "durationMs": 19.553500002075452,
      "charactersPerSecond": 2250.2365303055585,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 19.553500002075452,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3193.29591600399,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.26841699628858,
      "durationMs": 38.26841699628858,
      "charactersPerSecond": 1149.7731929770518,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3231.5643330002786,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 20.709374999569263,
      "durationMs": 20.709374999569263,
      "charactersPerSecond": 2124.6416176690586,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 20.709374999569263,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 20.285833001253195,
      "durationMs": 20.285833001253195,
      "charactersPerSecond": 2169.001391132512,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 20.285833001253195,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3247.609417005151,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 37.40724999806844,
      "durationMs": 37.40724999806844,
      "charactersPerSecond": 1176.2425733586936,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3285.0166670032195,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 19.01637500122888,
      "durationMs": 19.01637500122888,
      "charactersPerSecond": 2313.79534728131,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 19.01637500122888,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 18.203457999334205,
      "durationMs": 18.203457999334205,
      "charactersPerSecond": 2417.12316426963,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 18.203457999334205,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3183.0568329969537,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.18208300072001,
      "durationMs": 38.18208300072001,
      "charactersPerSecond": 1152.3729598296215,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3221.2389159976738,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 19.555042003048584,
      "durationMs": 19.555042003048584,
      "charactersPerSecond": 1534.1311972289843,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 19.555042003048584,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": true,
      "expected": []
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 19.107375002931803,
      "durationMs": 19.107375002931803,
      "charactersPerSecond": 2302.7757603149944,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 19.107375002931803,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 286.490042002697,
      "durationMs": 286.490042002697,
      "charactersPerSecond": 13250.024236319756,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 286.490042002697,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 19.530749996192753,
      "durationMs": 19.530749996192753,
      "charactersPerSecond": 2252.8576735955958,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 19.530749996192753,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 18.761833001917694,
      "durationMs": 18.761833001917694,
      "charactersPerSecond": 2345.1866347761784,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 18.761833001917694,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3236.618000002636,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 39.03158300090581,
      "durationMs": 39.03158300090581,
      "charactersPerSecond": 1127.2922238121596,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3275.649583003542,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 21.480915995198302,
      "durationMs": 21.480915995198302,
      "charactersPerSecond": 2048.3297830425604,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 21.480915995198302,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 20.344667005701922,
      "durationMs": 20.344667005701922,
      "charactersPerSecond": 2162.72893469666,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 20.344667005701922,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3223.530124996614,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 38.11449999921024,
      "durationMs": 38.11449999921024,
      "charactersPerSecond": 1154.416298283113,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3261.6446249958244,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 20.201416002237238,
      "durationMs": 20.201416002237238,
      "charactersPerSecond": 2178.0651413310406,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 20.201416002237238,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "cpu/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 19.69962500152178,
      "durationMs": 19.69962500152178,
      "charactersPerSecond": 2233.5450546191123,
      "activationPlacement": {
        "encoderOutput": "cpu"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 19.69962500152178,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3322.673084003327,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 307.22337499901187,
      "durationMs": 307.22337499901187,
      "charactersPerSecond": 143.21826911816694,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3629.896459002339,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 2,
          "end": 15
        },
        {
          "text": "Paris",
          "label": "location",
          "start": 24,
          "end": 29
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 1491.373749995546,
      "durationMs": 1491.373749995546,
      "charactersPerSecond": 20.115681934249945,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 30,
      "labelCount": 2,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102692352,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 1491.373749995546,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "moderate"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": true,
      "expected": []
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 226.7717079957947,
      "durationMs": 226.7717079957947,
      "charactersPerSecond": 194.02773118777208,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1111474176,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 226.7717079957947,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "start": 3752,
          "end": 3765,
          "text": "Alice Johnson",
          "label": "person"
        },
        {
          "start": 3775,
          "end": 3784,
          "text": "Microsoft",
          "label": "organization"
        },
        {
          "start": 3788,
          "end": 3795,
          "text": "Seattle",
          "label": "location"
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 1188.3932920027291,
      "durationMs": 1188.3932920027291,
      "charactersPerSecond": 3194.2287334884104,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 3796,
      "labelCount": 3,
      "chunks": 7,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1127088128,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 1188.3932920027291,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "moderate"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 40.50466700573452,
      "durationMs": 40.50466700573452,
      "charactersPerSecond": 1086.2945742467311,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1127088128,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 40.50466700573452,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 33.48341600212734,
      "durationMs": 33.48341600212734,
      "charactersPerSecond": 1314.0833658430936,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1127088128,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 33.48341600212734,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3314.74391699885,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 213.21212500333786,
      "durationMs": 213.21212500333786,
      "charactersPerSecond": 206.367250452155,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3527.956042002188,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 28.906334002385847,
      "durationMs": 28.906334002385847,
      "charactersPerSecond": 1522.1577387284171,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 28.906334002385847,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 24.007499996514525,
      "durationMs": 24.007499996514525,
      "charactersPerSecond": 1832.7605959132788,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 24.007499996514525,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3271.4647500033607,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 226.29620799853,
      "durationMs": 226.29620799853,
      "charactersPerSecond": 194.435427748245,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3497.7609580018907,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 26.99199999915436,
      "durationMs": 26.99199999915436,
      "charactersPerSecond": 1630.1126260143187,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 26.99199999915436,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "exactSpanLabelF1": 1,
      "minimumF1": 0.8,
      "validOffsets": true,
      "exactNegative": null,
      "expected": [
        {
          "text": "Alice Johnson",
          "label": "person",
          "start": 0,
          "end": 13
        },
        {
          "text": "Microsoft",
          "label": "organization",
          "start": 23,
          "end": 32
        },
        {
          "text": "Seattle",
          "label": "location",
          "start": 36,
          "end": 43
        }
      ]
    },
    "runtime": {
      "engine": "gliner",
      "version": "0.2.30",
      "transformersVersion": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "placement": {
        "mps:0/torch.float32": 128
      },
      "parameterDtype": "torch.float32",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "accelerationDetails": "DeBERTa encoder and GLiNER span parameters on selected device; tokenization, word splitting, decoding/offsets and JSON conversion use CPU",
      "modelMaxWords": 384,
      "maximumSpanWords": 12,
      "inferenceMs": 25.168207997921854,
      "durationMs": 25.168207997921854,
      "charactersPerSecond": 1748.2373001539518,
      "activationPlacement": {
        "encoderOutput": "mps:0"
      },
      "threshold": 0.5,
      "flatNer": true,
      "inputCharacters": 44,
      "labelCount": 3,
      "chunks": 1,
      "chunkWordBudget": 128,
      "chunkOverlapWords": 12,
      "inputTruncated": false,
      "offsetUnit": "Unicode code points, end-exclusive",
      "driverAllocatedMemoryBytes": 1102266368,
      "memoryMeasurementScope": "Post-inference Metal allocation, not peak",
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 25.168207997921854,
      "artifactRevision": "4e091416cf7c3481db542c2a3d26156916f3a47f",
      "fallback": false,
      "grade": "fast"
    }
  }
]

Something not quite right?

Tell us what happened with GLiNER 2.1. 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.