Induction Heads and In-Context Learning Mechanisms
Researchers trace in-context learning to a simple copying circuit in transformers.
Idris Petrova
Contributing Editor
Idris Petrova
Idris Petrova is a contributing editor at LLM Risk Review covering mechanistic. Based in Dublin, Idris has written for LLM Risk Review since 2017.
1 story · Dublin