concept
active
concept:sparse-circuits

Sparse circuits

A goal in mechanistic interpretability to identify sparse computational subgraphs; VPD promotes sparse parameter circuits.

Neighborhood — ranked by edge-count

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

  • Hypothesis that discretization finds minimum-size circuits equivalent to minimal algorithmic descriptions of patterns
  • Sparse dictionary learning variant that reads and writes to multiple layers or models, yielding shared features.
  • Default circuitsconcept0.780
    Circuits active by default, e.g. can't-answer, suppressed by contrary evidence.
  • Circuit Findingmethod0.776
    Interpretability technique for identifying functional sub-circuits in neural networks, supported by pyvene
  • Sparse subframeconcept0.774
    Mathematical object defining J-space as union of cones.
  • Sparse Crosscodersframework0.766
    Extension of SAEs that jointly learns latents across representations from different models; proposed as alternative for extended fine-tuning
  • The hypothesis that semantic concepts in neural representations can be captured using sparsity priors
  • Circuit tracingmethod0.764
    Anthropic technique revealing computational pathways in Claude.