Pinned Loading

1. RepoVeritas RepoVeritas Public

   A three-way code-grounding benchmark with an abstention (insufficient) class — "FEVER for code."

   Python 1

2. ntk-scaling-experiments ntk-scaling-experiments Public

   Independent reproduction and gate budget scaling experiments on NTK-based forward pass adaptation.

   HTML

3. ClinVeritas ClinVeritas Public

   Work in progress clinical note grounding dataset/protocol for Supported, Contradicted, and Insufficient transcript verification.

   Python 1

4. Planning-Navigator Planning-Navigator Public

   Python

5. SafePlant-Agent SafePlant-Agent Public

   Simulator gated AI fault diagnosis and action selection for chemical process control.

   Python 1

6. epistemic-gate epistemic-gate Public

   A preregistered stress test of hard pre-execution verification for tool agents.

   Python