framework
active
framework:dedicated-feature-crosscoder-dfcDedicated Feature Crosscoder (DFC)
Architectural variant partitioning the dictionary into A-exclusive, B-exclusive and shared sets.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (2)
method
- Dead-feature revival loss following Gao et al.
- Sparsity activation used with k=200.
Frameworks (2)
framework
- Baseline from Mishra-Sharma et al. dedicating features to shared variance.
- CrosscodersextendsLearn a shared feature space across two models; basis of the work.
Datasets (1)
dataset
- LMSYS-CHAT-1McitesChat dataset (n=18,777 sampled) used to measure how much persona space PCs explain overall activation variance and to calibrate steering norms
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.
- Main architectural claim.
- Table 1.
- Figure 5.
- Hyperparameter sweep.
- Figure 4 left.
- Motivation for DFC.
- Source of the features used in attributions.
- Qualitative comparison.