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.

RuBERT Toxicity

Explore text labels and scores, processed locally in your browser.

Browser inferencePowered by WebAI · Private · Unlimited runs
About this model

Flag potentially toxic or inappropriate short Russian text. Returns scores for five overlapping categories and an aggregate toxicity score; use these predictions to assist review, not as an automatic final judgment.

Created by
cointegrated
License
MIT
Model released
Unknown
Source repository created
Mar 2, 2022
Model weights · FP32
47.2 MB

Best for

  • Screen short Russian comments for toxicity signals.
  • Inspect insult, obscenity, threat, and reputational-danger scores.
  • Benchmark local browser text classification across ONNX variants.

Keep in mind

  • Designed for short informal Russian text; other languages, long documents, dialects, coded language, and domain-specific usage may be unreliable.
  • Scores can encode dataset bias and can produce both false positives and false negatives.
  • The dangerous label concerns potential reputational harm and is not interchangeable with toxicity.
  • Automated scores require human review before moderation or other consequential decisions.
  • The model does not understand speaker intent or broader conversational context.

Processed on your deviceInference runs in this browser. Input/output sharing requires consent for each run. Basic run logs include device information and approximate connection location. Model and runtime files download from external hosts. Saved downloads may be cleared by your browser.

Sources & technical details
Original model
cointegrated/rubert-tiny-toxicity
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
Aug 3, 2026
Browser model files
onnx-community/rubert-tiny-toxicity-ONNX
Files updated
Sep 9, 2026
Artifact revision
3c117322c91030ff04846b72e15509ae7f5232f1

Input

TEXT → LABELS

Sample result

RuBERT Tiny Russian Toxicity
Advanced settings 1
Maximum token count after truncation. Lower values may process less text but truncate relevant context sooner. Higher values retain more context, within the model limit, and can need more memory and time.

Output

Explore your text’s labels

Enter your text or try the sample to see predicted labels.

Benchmark evidence

Saved worker evidence is separate from benchmarks run on your device. It does not guarantee speed or quality on every browser.

RuBERT Tiny Russian Toxicity

12 configurations tested
Passed12
Failed0
TestedSep 15, 2026
RuntimeTransformers.js 4.2.0

Quality criteria

The sourced toxic and safe examples must cross their respective aggregate-toxicity thresholds; every score must be finite and within zero to one.

PrecisionDeviceResult
Q4GPU · WebGPUAcceleratedPassedVery fast
FP32GPU · WebGPUAcceleratedPassedVery fast
BNB4GPU · WebGPUAcceleratedPassedVery fast
INT8GPU · WebGPUAcceleratedPassedVery fast
Q8GPU · WebGPUAcceleratedPassedFast
UINT8GPU · WebGPUAcceleratedPassedFast
FP32CPU · WASMPassedVery fast
INT8CPU · WASMPassedVery fast
Q8CPU · WASMPassedVery fast
Q4CPU · WASMPassedVery fast
BNB4CPU · WASMPassedVery fast
UINT8CPU · WASMPassedVery fast
Test setup & sample details
Quality gate
The sourced toxic and safe examples must cross their respective aggregate-toxicity thresholds; every score must be finite and within zero to one.
Tested
2026-09-15T13:09:47.219Z
Environment
Isolated headless Chrome on macOS; Transformers.js 4.2.0
Runtime
Transformers.js 4.2.0
Artifact revision
3c117322c91030ff04846b72e15509ae7f5232f1
Fixture
{
  "file": "rubert-tiny-toxicity.json",
  "url": "https://f005.backblazeb2.com/file/hubters-webai-public/assets/benchmarks/rubert-tiny-toxicity.json",
  "size": 700,
  "sha1": "f136212c8f17ea8b87f6c567580ddae3ad924bcc",
  "source": "Russian examples from the pinned source model card and widget metadata.",
  "sourceUrl": "https://huggingface.co/cointegrated/rubert-tiny-toxicity/blob/5d37eff844868e243467e4c38898bad46c271af2/README.md",
  "license": "MIT",
  "licenseUrl": "https://huggingface.co/cointegrated/rubert-tiny-toxicity/blob/5d37eff844868e243467e4c38898bad46c271af2/README.md",
  "preprocessing": "Pinned tokenizer truncates each Russian string to max_length."
}
Rerun when
artifact revision change; runtime version change; fixture or quality-gate change; browser/backend behavior change
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 12711,
      "inferenceMs": 23.5,
      "totalMs": 13994.5,
      "textsPerSecond": 85.1063829787234,
      "grade": "excellent"
    }
  },
  {
    "precision": "fp32",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 10926.9,
      "inferenceMs": 67.5,
      "totalMs": 12316.299999982119,
      "textsPerSecond": 29.62962962962963,
      "grade": "excellent"
    }
  },
  {
    "precision": "int8",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 7104.7,
      "inferenceMs": 27,
      "totalMs": 8447.59999999404,
      "textsPerSecond": 74.07407407407408,
      "grade": "excellent"
    }
  },
  {
    "precision": "int8",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 6963.3,
      "inferenceMs": 92.1,
      "totalMs": 8382.09999999404,
      "textsPerSecond": 21.715526595898602,
      "grade": "excellent"
    }
  },
  {
    "precision": "uint8",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 7128.8,
      "inferenceMs": 30.2,
      "totalMs": 8386.59999999404,
      "textsPerSecond": 66.22516558905515,
      "grade": "excellent"
    }
  },
  {
    "precision": "uint8",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 7330.5,
      "inferenceMs": 104.9,
      "totalMs": 8656.799999982119,
      "textsPerSecond": 19.065776934743216,
      "grade": "good"
    }
  },
  {
    "precision": "q8",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 6940.3,
      "inferenceMs": 27.7,
      "totalMs": 8171.299999982119,
      "textsPerSecond": 72.20216601837274,
      "grade": "excellent"
    }
  },
  {
    "precision": "q8",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 7364.6,
      "inferenceMs": 103.8,
      "totalMs": 8777.299999982119,
      "textsPerSecond": 19.267822739350052,
      "grade": "good"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 9238.1,
      "inferenceMs": 28.2,
      "totalMs": 10615.399999976158,
      "textsPerSecond": 70.92198577063178,
      "grade": "excellent"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 9954.1,
      "inferenceMs": 65.8,
      "totalMs": 11337.59999999404,
      "textsPerSecond": 30.39513677260884,
      "grade": "excellent"
    }
  },
  {
    "precision": "bnb4",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 9791.5,
      "inferenceMs": 29.4,
      "totalMs": 11250.299999982119,
      "textsPerSecond": 68.02721093952024,
      "grade": "excellent"
    }
  },
  {
    "precision": "bnb4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "grade": "toxicity-regression-pass"
    },
    "runtime": {
      "loadMs": 11344.1,
      "inferenceMs": 88.4,
      "totalMs": 12691.300000011921,
      "textsPerSecond": 22.624434387614794,
      "grade": "excellent"
    }
  }
]

Something not quite right?

Tell us what happened with RuBERT Toxicity. 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.