Quantized ONNX export of FacebookAI/roberta-large-mnli at commit 2a8f12d27941090092df78e4ba6f0928eb5eac98. Hosted as the canonical Layer 2 model for bh-sentinel-ml ≥ 0.2.2.

Clinical Use Notice

This is clinical decision support software. It is not a diagnostic tool, not FDA-cleared, and not a substitute for clinical judgment. The bh-sentinel pipeline that consumes this artifact is intended only for flagging signals for clinician review — never for autonomous clinical action. All outputs are signals for clinician review. See the main repository's clinical disclaimer for the full notice.

License Chain

Layer	Model	License	Source
Pretraining base	`FacebookAI/roberta-large`	MIT	Facebook AI Research (Meta Platforms, Inc.)
Source (MNLI fine-tune)	`FacebookAI/roberta-large-mnli`	MIT	Facebook AI Research
This artifact	`bh-healthcare/roberta-large-mnli-int8-onnx`	MIT	bh-healthcare (this repo)
Consuming code	`bh-sentinel-ml`	Apache-2.0	bh-healthcare

The redistribution is permitted under the upstream MIT terms. The original copyright notice is preserved verbatim in LICENSE. The Python package that loads this artifact is Apache-2.0 — different copyrightable works can carry different terms; this is a common open-source pattern and is not contradictory.

Provenance

Field	Value
Source model	`FacebookAI/roberta-large-mnli`
Source revision	`2a8f12d27941090092df78e4ba6f0928eb5eac98`
Source paper	Liu et al. 2019, RoBERTa: A Robustly Optimized BERT Pretraining Approach
NLI fine-tune dataset	MultiNLI
Zero-shot method paper	Yin et al. 2019
Export tool	`optimum.exporters.onnx.main_export()` (via `scripts/export_onnx.py`'s Python API path)
`optimum` version	`2.1.0`
`onnxruntime` version	`1.26.0`
`onnx` version	`1.21.0`
Quantization	`onnxruntime.quantization.quantize_dynamic(weight_type=QInt8, per_channel=True, reduce_range=False)`
Export date (UTC)	`2026-05-14`
Export script	`scripts/export_onnx.py` at tag `ml-v0.2.2`

To reproduce locally, install optimum[onnxruntime]>=2.0,<3 and onnx>=1.15 and run the export script against the source revision above. The artifact is fully reproducible from those inputs.

Why RoBERTa-large-MNLI and not BART-large-MNLI

A prior iteration of this artifact pinned facebook/bart-large-mnli. BART is encoder-decoder; under INT8 dynamic quantization the classification head's discrimination collapsed (FP32 emitted entailment probabilities near 0.998 on clear positive cases, INT8 emitted near 0.24 with similar magnitudes for negative controls — effectively noise). Encoder-only models like RoBERTa quantize cleanly: this artifact preserves FP32 discrimination to within <1% on the same test cases. See the bh-sentinel repository's docs/ml-artifact-provenance.md for the full source-selection narrative.

Size & Performance

Metric	Value	Notes
ONNX file size (INT8)	`342` MB	Down from ~1.4GB FP32
SHA256 of `model_int8.onnx`	`49fa5562da7f1422525e88fd1145d2d06ca93b17c335adf3c4696a30908c91de`	Pinned in `bh-sentinel-ml`'s `ml_config.yaml`; verified at load time
First-call download time (typical broadband)	~20–60s	One-time; subsequent calls hit the local platformdirs cache
Per-sentence inference latency (CPU, single thread)	~40–80 ms	Per-pair (premise, hypothesis) NLI forward pass; encoder-only is faster than encoder-decoder
Per-flag inference cost	~40–80 ms × N_hypotheses_per_flag	bh-sentinel-ml batches these; see `max_batch_size` in `ml_config.yaml`
Recommended Lambda memory	1.5GB+	INT8 model + onnxruntime CPU provider + tokenizer cache
Recommended Lambda image size budget	600MB+	Model baked in; `BH_SENTINEL_ML_OFFLINE=1` Dockerfile pattern

These numbers are operator-reported approximations for capacity planning. Real-world latency varies with CPU class, ARM vs x86, ONNX Runtime threading config, and concurrent load. The bh-sentinel-ml pipeline supports BH_SENTINEL_ML_OFFLINE=1 for fully air-gapped VPC deployments where the model is pre-baked into the container image.

Files

File	Purpose
`model_int8.onnx`	Quantized model weights, `342` MB. SHA256 `49fa5562da7f1422525e88fd1145d2d06ca93b17c335adf3c4696a30908c91de` (consumed by `TransformerClassifier`'s verify-on-load check)
`tokenizer.json`	HF `tokenizers` library serialized BPE tokenizer (required by `Tokenizer.from_file()` in the runtime)
`tokenizer_config.json`	Tokenizer config (model_max_length, special tokens)
`vocab.json`	BPE vocabulary
`merges.txt`	BPE merge rules
`special_tokens_map.json`	Special token map (BOS, EOS, PAD, etc.)
`manifest.json`	Machine-readable provenance metadata (matches the values in the Provenance section above)
`LICENSE`	MIT license text + attribution to Facebook AI Research / Meta Platforms, Inc.

Intended Use

Consumed by bh-sentinel-ml ≥ 0.2.2 via Pipeline(enable_transformer=True). The bh-sentinel pipeline performs zero-shot NLI inference against a curated set of clinical hypothesis templates (see config/ml/zero_shot_hypotheses.yaml) to surface candidate clinical safety flags.

Use cases this artifact is designed for:

Layer 2 (semantic) detection in the bh-sentinel multi-layer pipeline
Catching clinical signals that pattern-matching alone misses: indirect language, implied distress, contextual meaning
Reproducible local evaluation of bh-sentinel against the shared corpus in bh-sentinel-examples

Out-of-Scope Use

This artifact is not appropriate for:

Direct clinical diagnosis or treatment decisions
Autonomous clinical action of any kind
Use as the sole basis for any clinical decision
Inference tasks other than zero-shot NLI (it has not been retrained for other tasks)
Production deployments without independent clinical validation against your organization's de-identified data
Fine-tuning starting points for high-stakes clinical applications without further validation

Limitations

Trained on general-purpose MNLI, not clinical text. Confidence calibration on clinical inputs is dampened via FixedDiscount(0.85) in bh-sentinel-ml's Phase A calibrator, but is not clinically validated. Validated calibration against clinician-labeled data is a bh-sentinel v0.3 deliverable.
INT8 dynamic quantization introduces small numerical drift relative to the FP32 source. Spot-check shows <1% drift on representative NLI cases; the bh-sentinel test suite includes a real-model L2 smoke test that asserts the discrimination is preserved end-to-end.
English only. The MNLI training data is English-language; behavior on other languages is undefined and unsafe.
No clinical labels in training. The model has never seen clinician-labeled examples of behavioral health safety signals. Its outputs are best understood as plausibility scores against pre-defined hypothesis templates, not as clinical probabilities.

Citation

If you use this artifact, please cite both the upstream source and bh-sentinel:

bibtex

@misc{roberta-large-mnli,
  title  = {roberta-large-mnli},
  author = {Facebook AI Research},
  year   = {2019},
  publisher = {Hugging Face},
  url    = {https://huggingface.co/FacebookAI/roberta-large-mnli}
}

@article{liu2019roberta,
  title   = {{RoBERTa}: A Robustly Optimized {BERT} Pretraining Approach},
  author  = {Liu, Yinhan and Ott, Myle and Goyal, Naman and Du, Jingfei and Joshi, Mandar and Chen, Danqi and Levy, Omer and Lewis, Mike and Zettlemoyer, Luke and Stoyanov, Veselin},
  journal = {arXiv preprint arXiv:1907.11692},
  year    = {2019},
  url     = {https://arxiv.org/abs/1907.11692}
}

@article{yin2019benchmarking,
  title   = {Benchmarking Zero-shot Text Classification: Datasets, Evaluation and Entailment Approach},
  author  = {Yin, Wenpeng and Hay, Jamaal and Roth, Dan},
  journal = {arXiv preprint arXiv:1909.00161},
  year    = {2019},
  url     = {https://arxiv.org/abs/1909.00161}
}

@software{bh_sentinel,
  title  = {bh-sentinel: Open-source clinical safety signal detection for behavioral health systems},
  year   = {2026},
  publisher = {bh-healthcare},
  url    = {https://github.com/bh-healthcare/bh-sentinel}
}

Acknowledgments

The source model FacebookAI/roberta-large-mnli was authored by Facebook AI Research (now Meta AI) and is redistributed here under its original MIT license.
The zero-shot NLI classification method was proposed by Yin et al. 2019; HF's blog post by Joe Davison (May 2020) popularized the pattern this artifact serves.
The optimum library by Hugging Face provides the ONNX export path used here.

bh-healthcare/roberta-large-mnli-int8-onnx

RoBERTa-Large-MNLI INT8 ONNX (bh-sentinel pinned artifact)