method
active
method:pinned-feature-samplingPinned Feature Sampling
Setting a feature's value to its maximum observed value and sampling from the model to validate causal interpretations
Neighborhood — ranked by edge-count
Claims (1)
claim
- Authors argue features are model properties because logit effects and ablations are consistent with feature interpretations
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.
- The mechanism by which LLMs generate text: drawing a token from the next-token distribution and appending it to context repeatedly
- A technique to filter model outputs; Redwood Research's project mentioned.
- Method of optimizing input to cause a neuron to fire maximally, used to characterize what a neuron detects; establishes causal link
- Repetitive behavioral pattern observed under high steering strengths in SAE feature self-steering experiments
- Temperature=0.8 sampled decoding for self-report; reduces collapse moderately but remains discrete and noisy
- Examining downstream neurons that rely on a given feature to verify its functional role
- Property that features activate on only a small fraction of inputs; enables compressed sensing and is what allows superposition to work
- Dividing feature activation spectrum into 11 evenly-spaced intervals and sampling uniformly to evaluate monosemanticity across activation levels