method
active
method:transcoderTranscoder
Uses activations from one layer to predict the next; special case of crosscoders.
Neighborhood — ranked by edge-count
Methods (3)
method
- Transcodersrelated_toDecomposition method for activations; VPD is compared against transcoders in sparsity-reconstruction tradeoff.
- Single-layer transcoderrelated_toUsed to express MLP in monosemantic features, 4096 features with JumpReLU.
- Sparse CrosscoderextendsSparse dictionary learning variant that reads and writes to multiple layers or models, yielding shared features.
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.
- MLP replacement with sparse features underlying the original attribution graphs.
- Learn a shared feature space across two models; basis of the work.
- Decomposition approach mixing frozen base attention patterns.
- Second category of SAE/transcoder concerns, distinct from faithfulness.
- Mask virtual weights during training targeting interference.
- Neural network architecture that learns compressed representations; SOHMs are functionally equivalent.
- Replacement model architecture with 30M features replacing MLP neurons.