finding
active
finding:repeated-echo-features-appear-in-the-llama-3-8b-saeRepeated echo features appear in the Llama-3-8B SAE
Seen as in other SAEs.
Source paper
extracted_fromNeighborhood — ranked by edge-count
Papers (1)
paper
Hypotheses (1)
hypothesis
- Mitigation proposal.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Moderation statistics.
- SAE reconstructions on Llama-3-8B layer 25 produce intervened EMD exceeding the natural-natural baselinefinding0.782Empirical demonstration that SAE projections produce divergent representations in a real LLM
- App. E.
- Llama-3.3-70B exhibits internal consistency-checking mechanisms that operate during inferenceclaim0.753Central interpretive claim of the paper supported by causal ablation and activation evidence
- One of four LLMs selected for representation analysis; embedding dimension D=4096; used as demonstration model in scatter plots.
- Goodfire blog post describing SAEs used for Llama models in this study
- Open-source SAEs from Goodfire used for Llama model experiments, trained on instruction-tuned models
- Beliefs linearly decodable.