Verified75% confidenceFactExact time
Anthropic published mechanistic interpretability research on the internal feature interpretation of Claude in 2024, discovering millions of interpretable feature units.
5
Sources
75%
Confidence
Long-term
Relevance
8/4/2026
First Seen
Sources
Related Claims
UnverifiedBetween 2023 and 2024, Anthropic identified millions of interpretable features in the Claude model, including the Golden Gate Bridge feature experiment79% similarPartially VerifiedIn 2024, Anthropic's mechanistic interpretability team used sparse autoencoders to extract millions of interpretable concept units from the Claude model.77% similarUnverifiedIn 2024, Anthropic successfully decomposed millions of interpretable features in the Claude model using sparse autoencoders (SAE).76% similarUnverifiedAnthropic于2023年发表论文《Towards Monosemanticity》，首次在百万参数级别模型中识别出数万个单语义特征76% similarUnverifiedIn a paper published by Anthropic in 2023, sparse autoencoders were used to find tens of thousands of interpretable features in Claude, including emotional states and programming language types.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/684191API
curl https://kongchang.com/api/v1/knowledge/claims/684191MCP
get_claim(id=684191)