framework
active
framework:multi-token-transcodersMulti-token transcoders
Decomposition approach mixing frozen base attention patterns.
Neighborhood — ranked by edge-count
Papers (2)
paper
Methods (1)
method
- Strategy forcing edges to be mediated only by head paths of length 1 to avoid exponential path growth.
Frameworks (1)
framework
- Low-Rank Sparse Attention (Lorsa)associated_withLearns new full-rank QK circuits shared across feature groups.
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.
- Decomposition method for activations; VPD is compared against transcoders in sparsity-reconstruction tradeoff.
- Uses activations from one layer to predict the next; special case of crosscoders.
- Used to express MLP in monosemantic features, 4096 features with JumpReLU.
- MLP replacement with sparse features underlying the original attribution graphs.
- Mask virtual weights during training targeting interference.
- Replacement model architecture with 30M features replacing MLP neurons.
- Learn a shared feature space across two models; basis of the work.
- Second category of SAE/transcoder concerns, distinct from faithfulness.