Re-derivation of activation patching NIE reveals it captures interaction effects in addition to direct causal effects, demonstrated via GPT-2 IOI circuit where INT explains component ranking issues and faithfulness instability.
Computational Linguistics , pages =
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
years
2026 5verdicts
UNVERDICTED 5representative citing papers
Manifold steering along activation geometry induces behavioral trajectories matching the natural manifold of outputs, while linear steering produces off-manifold unnatural behaviors.
Llama-3.1-8B computes sums for cyclic concepts using base-10 addition via task-agnostic Fourier features with periods 2, 5, and 10 rather than modular arithmetic in the concept period.
Linear probes recover day-of-year from LM activations for temporal reasoning but are orthogonal to the model's causal 4D subspace identified by DAS, with the angle matching the Haar-uniform random null, replicated across scales and families.
Instructions trigger a production-centered mechanism in language models, with task-specific information stable in input tokens but varying strongly in output tokens and correlating with behavior.
citing papers explorer
-
The Curse of Multiple Mediators: Hidden Interaction Effects in Activation Patching
Re-derivation of activation patching NIE reveals it captures interaction effects in addition to direct causal effects, demonstrated via GPT-2 IOI circuit where INT explains component ranking issues and faithfulness instability.
-
Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior
Manifold steering along activation geometry induces behavioral trajectories matching the natural manifold of outputs, while linear steering produces off-manifold unnatural behaviors.
-
Arithmetic in the Wild: Llama uses Base-10 Addition to Reason About Cyclic Concepts
Llama-3.1-8B computes sums for cyclic concepts using base-10 addition via task-agnostic Fourier features with periods 2, 5, and 10 rather than modular arithmetic in the concept period.
-
When and How Long? The Readout-Mediator Angle in Temporal Reasoning
Linear probes recover day-of-year from LM activations for temporal reasoning but are orthogonal to the model's causal 4D subspace identified by DAS, with the angle matching the Haar-uniform random null, replicated across scales and families.
-
Instructions Shape Production of Language, not Processing
Instructions trigger a production-centered mechanism in language models, with task-specific information stable in input tokens but varying strongly in output tokens and correlating with behavior.