Skip to main contentSkip to footer
Try Claude
InterpretabilityResearch

In-context Learning and Induction Heads

Mar 8, 2022
Read Paper

Related content

Estimating AI productivity gains from Claude conversations

Read more

Mitigating the risk of prompt injections in browser use

Read more

From shortcuts to sabotage: natural emergent misalignment from reward hacking

Read more
In-context Learning and Induction Heads \ Anthropic