framework
active
framework:per-token-saePer-Token SAE
Standard baseline SAE operating on individual token activations.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (2)
method
- Alternative aggregation giving high discrimination.
- Aggregation of token-level activations.
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.
- Table 1.
- Example A interpretation.
- Motivation.
- Interpretable features extracted by sparse autoencoders used as steering targets in this study
- Table 1.
- The specific SAE architecture trained: 100K+ possible features compressed to 64 active per token for layer-40 activations
- Table 1.
- Pre-filtering step excluding latents naturally activated by each prompt to ensure genuine off-topic steering