method
active
method:auxk-auxiliary-lossAuxK auxiliary loss
Dead-feature revival loss following Gao et al.
Neighborhood — ranked by edge-count
Papers (1)
paper
Frameworks (1)
framework
- Architectural variant partitioning the dictionary into A-exclusive, B-exclusive and shared sets.
Related by similarity (6)
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.
- Auxiliary objective combining L2 and cosine losses against pre-recorded CL vectors to improve causal relevance when one model is causally inaccessible.
- In machine learning, a function measuring the distance between current and desired output; analogous to stress.
- Reconstruction fidelity score based on cross-entropy.
- Regularization component of the composite loss that penalizes deviation from baseline model behavior on Alpaca instructions
- Loss on randomly masked subcomponents matching target outputs, with layerwise auxiliary.
- Novel variant of CL loss introduced in this paper targeting only causal subspace dimensions to improve OOD performance