Verifiable circuit tasks: how training data for circuit interpretability could look
yoavgo · x · 2026-11-03
On what training data for circuit-level tasks would look like, Goldberg suggests treating them like any other verifiable task: "here's a circuit, explain which of these 4 options it implements" or "write a circuit capable of performing these operations" — tasks with automatically checkable answers.
More from Research
- LocalFold update brings BioEmu2, Mol-1, AF3x crosslinks to browser-based protein folding — CatAstro_Piyush · 2026-11-03
- kalomaze conjectures deep sequence models could mimic restricted training families like rank-1 attention LoRAs — kalomaze · 2026-11-03
- Zyphra's ZHA brings subquadratic attention via hierarchical tree-structured indexer — yacineMTB · 2026-11-03
- kalomaze: In-context learning is an incidental byproduct of next-token prediction, not deliberate design — kalomaze · 2026-11-03
- RSNA 2026 lab walks through fine-tuning a VLM for radiology end to end — PRouzrokh · 2026-11-03
- RSNA 2026 lab walks through fine-tuning a VLM for radiology end to end — PRouzrokh · 2026-11-03