finding
active
finding:inception-v1-sae-shows-meaningful-fraction-of-latents-with-very-high-nearest-neighbor-cosineInception-v1 SAE shows meaningful fraction of latents with very high nearest-neighbor cosine
Fig 8; may be duplication.
Source paper
extracted_from(2025) · Michaud, Eric J. · Gorton, Liv · McGrath, Tom
Neighborhood — ranked by edge-count
Papers (1)
paper
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.
- Training artifact.
- Fig 2 right.
- Model diffing identifies a small, interpretable set of latents responsible for emergent misalignment
- Claim that feature grounding enables interpretability metrics.
- The most decreased latent after bad-advice fine-tuning is also the most effective re-aligning latent
- SAE features are not simply mirroring individual neurons.
- Latent #10 activation increase correctly classifies all correct vs incorrect fine-tuned models in Figure 9
- SAEs assume sparsity (number of concepts) is time-invariant, drawn i.i.d. from a fixed distribution.claim0.765Corollary 4.1.1.