Hf Emilyalsentzer Bio Clinicalbert

This Biosimulant lab wraps a source-derived ONNX artifact and reports runtime evidence from the bundled graph.

Scientific Context

ClinicalBERT clinical-text embedding exported to ONNX.

Source And Artifact

• model: https://huggingface.co/emilyalsentzer/Bio_ClinicalBERT

• ONNX artifact: models/core/artifacts/model.onnx

• Runtime: ONNX Runtime CPU execution provider

• Validation scope: graph load, input/output contract, finite synthetic inference, Biosimulant wiring, and visual payloads

Inputs

• clinical_text_token_ids: Inputs map to ONNX input_ids and attention_mask; tokenizer preprocessing is not bundled.
• clinical_text_attention_mask: Inputs map to ONNX input_ids and attention_mask; tokenizer preprocessing is not bundled.

The lab does not add raw image, raw text, raw protein sequence, tokenizer, or medical-image preprocessing unless that preprocessing is bundled and validated. Public inputs are already-preprocessed tensors matching the ONNX graph contract.

Outputs

• sequence_embeddings: Compact embedding evidence derived from the source ONNX graph output.
• pooled_embedding: Compact embedding evidence derived from the source ONNX graph output.

Visualisations

The visualisation answers:

What embedding tensors does the ClinicalBERT ONNX export emit for these already-tokenized clinical text inputs?

It shows provenance, graph schema, runtime tensor evidence, and a conservative caveat. Synthetic inference is a runtime smoke test, not biological, clinical, or microscopy performance evidence.

Caveat

No tokenizer is bundled, so public inputs are token tensors rather than raw text.

{
  "title": "Hf Emilyalsentzer Bio Clinicalbert Lab",
  "models": [
    {
      "path": "models/core",
      "alias": "hf_emilyalsentzer_bio_clinicalbert"
    },
    {
      "path": "models/visualisation",
      "alias": "visualisation"
    }
  ],
  "wiring": [
    {
      "to": [
        "visualisation.sequence_embeddings"
      ],
      "from": "hf_emilyalsentzer_bio_clinicalbert.sequence_embeddings"
    },
    {
      "to": [
        "visualisation.pooled_embedding"
      ],
      "from": "hf_emilyalsentzer_bio_clinicalbert.pooled_embedding"
    }
  ],
  "package": "hf-emilyalsentzer-bio-clinicalbert",
  "runtime": {
    "duration": 0.02,
    "initial_inputs": {},
    "communication_step": 0.01
  },
  "version": "1.0.0",
  "description": "ClinicalBERT clinical-text embedding exported to ONNX. This lab exposes source-faithful ONNX runtime evidence without claiming scientific validity beyond the source artifact.",
  "schema_version": "2.0"
}