finding
active
finding:baseline-jacobian-on-the-memorized-point-is-diagonal-while-with-datapoint-feature-each-feature-s-activity-depends-on-all-othersBaseline Jacobian on the memorized point is diagonal, while with datapoint feature each feature's activity depends on all others
Motivates Jacobian connection.
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Rationale for Jacobian matching.
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.
- Adding Jacobian matching loss dramatically reduces datapoint-feature memorization metricfinding0.816Result of Jacobian matching.
- Toy-model fix of unfaithfulness.
- Hyperparameter scan result.
- Datapoint features form only if there is a repeated datapoint in training and sparsity regularization is appliedhypothesis0.770Two necessary conditions.
- Demonstrates unfaithfulness in a simple setup.
- Memorization by transcoder not in model.
- Quantitative relationship between concept frequency and feature presence.
- Ignition-like effect.