concept
active
concept:probesProbes
Interpretability tools that decode information from internal model activations; here, linear probes are used for data attribution.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (2)
method
- Linear ProbeimplementsSimple linear classifiers trained on model activations used as the probing technique within the introduced method.
- Linear classifier approach applied to model activations to identify which training datapoints caused undesired behaviors in post-training.
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.
- Attention probe predicting whether an Entity is hallucinated; AUC .94 on validation.
- Standard interpretability technique used to read feature beliefs.
- Internals-based approach to which the logit monitor is complementary.
- Causal transformer probe predicting whether a token is in the same Entity as its predecessor.
- Earlier interpretability method applying classifiers to DNN hidden representations; shares complexity-accuracy dilemma with causal abstraction
- Top-down interpretability approach studying linguistic properties at various residual stream stages; contrasted with the paper's bottom-up mechanistic approach
- Joint firing raises TPR at slightly higher FPR.
- Dot product between hidden state and concept vector averaged across 5-layer window around best layer; measures model's internal emotive state