method
active
method:amplified-logits-samplingAmplified logits sampling
Iterative sampling procedure recomputing before/after logits with each new token.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (1)
method
- Core method: sample from logits_after + α(logits_after − logits_before) to surface rare behaviors.
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.
- Prior method extrapolating beyond the original model in logit space to surface rare behaviors.
- Dot product of summed decoder vector with unembedding vector, following Hanna et al.
- The mechanism by which LLMs generate text: drawing a token from the next-token distribution and appending it to context repeatedly
- Samples from interpolated logits of original and abliterated models at α.
- Summary claim.
- Dividing feature activation spectrum into 11 evenly-spaced intervals and sampling uniformly to evaluate monosemanticity across activation levels
- Fits log compliance vs α on α with compliance <50% and extrapolates to α=1.
- Problem cited as a limitation of current LLMs; PRH predicts larger models should amplify bias less