concept
archived
concept:addition-moduleAddition module
Specific neural circuitry in layer 18 of Llama-3.1-8B that performs arithmetic and cyclic reasoning.
Neighborhood — ranked by edge-count
Papers (1)
paper
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.
- The mathematically natural computation for cyclic concepts (e.g., addition mod 12 for months), which the paper shows Llama does NOT directly implement
- Intervention method that adds a learned direction vector to residual stream activations to steer model behavior
- The generic addition mechanism that Llama-3.1-8B actually uses to compute sums before mapping back to cyclic concept space
- Steering method deriving vectors from contrastive prompt pairs and adding to first-token activations.
- The core finding: Llama-3.1-8B reuses one addition mechanism across all cyclic tasks rather than learning task-specific modular arithmetic
- Independent, parallel cognitive subsystems that share information via a global workspace.
- Adding steering vector in forward direction to push model activations toward stronger reflective behavior.
- Verbatim summary of the main discovery.