Honesty.Tools catalogs AI interpretability instruments by research task
kevrussell · x · 2026-11-04
Honesty.Tools, an independent research observatory, catalogs interpretability and AI-safety research instruments by task: inspecting representations, comparing and transporting, testing causal interpretations, calibrating measurements, studying feedback, and auditing evidence. Each entry lists method, output, limits, maturity, access, and reproducibility boundaries — some tools are publicly released, others are research prototypes or source-unreleased. A practical index for mechanistic interpretability researchers.
More from Research
- RSI-Dojo: Training LLM Agents to Automate Post-Training, Matching Frontier Models — dair_ai · 2026-11-04
- Replicating Anthropic's 'Teaching Claude Why': hard advice cuts misalignment to 0.04 — dhadfieldmenell · 2026-11-04
- Formalizing PTX in Lean catches kernel bugs that testing misses — mishig25 · 2026-11-04
- The Trinity couples verifiable world generation with agent harness evolution, gaining +26 on unseen benchmarks — YangYou1991 · 2026-11-04
- Editing 0.13% of weights, no training, fixes LightOnOCR-2-1B blank-page hallucination — vanstriendaniel · 2026-11-04
- antirez: benchmark games trivially guessable from output biases; Astra misses Cannon — antirez · 2026-11-04