method
active
method:interchange-intervention-training-iitInterchange Intervention Training (IIT)
Training technique that induces specific causal structures in neural networks by co-training with interchange interventions
Neighborhood — ranked by edge-count
Papers (1)
paper
Concepts (1)
concept
- Causal abstractionimplementsA framework the paper uses alongside feature geometry to deepen mechanistic understanding of LMs
Artifacts (1)
artifact
- pyvene open-source Python libraryimplementsThe main artifact introduced in the paper: an open-source PyPI library for customizable interventions on PyTorch models
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.
- Differentiable training objective minimized when a high-level model is an abstraction of a neural network under a given alignment.
- Evaluation metric measuring how well a trained intervention matches desired counterfactual model behavior
- Metric measuring accuracy of DNN under intervention at matching algorithm-predicted outputs on held-out test set
- Fundamental operation for causal abstraction analysis; forces neurons to take values from source inputs to create counterfactuals.
- Proportion of aligned interchange interventions with equivalent high-level and low-level effects; graded measure of causal abstraction.
- Extends interchange interventions to non-standard bases by rotating representations, intervening in rotated subspaces, then rotating back.
- Version 3.0 of IIT, used to compute Φmax and Conceptual Information (CI) from LLM representation networks.
- Version 4.0 of IIT, used to compute Φ and Φ-structure from LLM representation networks; latest iteration at time of study.