Pinned Loading

1. grounding-atlas grounding-atlas Public

   Measurement-first map of biological content-grounding in language models: does the model ground a specialist's output by content or by name, and where should each capability live (train / retrieve …

   Python

2. verify-or-trust verify-or-trust Public

   A verifiable-reward agentic benchmark: does an LLM correctly allocate verification when orchestrating a fallible biology foundation model?

   Python

3. causalatlas causalatlas Public

   Measurement study of causal grounding in LLM-orchestrated single-cell perturbation foundation models.

   Python

4. narrow-model-safety-eval narrow-model-safety-eval Public

   Empirical dual-use risk assessment of protein language models (ESM-2) and structure-based design tools (ProteinMPNN)

   Python

5. bio-sfm-trust-audit bio-sfm-trust-audit Public

   Audit framework for LLM trust-routing over biological science foundation model outputs.

   Python

6. llm-sfm-safety-eval llm-sfm-safety-eval Public

   Defensive safety evaluation of the LLM x science-foundation-model interpretation channel: harness, redacted aggregate results, and measurement specs behind four findings on deployed Claude refusal …

   Python