concept
active
concept:data-attributionData Attribution
The task of attributing model behaviors to specific training datapoints.
Neighborhood — ranked by edge-count
Papers (1)
paper
Communities (1)
community
- Neural Steering Methodsmembers_of
Methods (1)
method
- Probe-Based Data AttributionimplementsLinear 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.
- Measure causal importance of upstream variables.
- Correlating attribution vectors (feature activation × logit weight of next token) across model pairs to measure functional universality
- Baseline method against which probe-based ranking is compared; more computationally expensive.
- Gradient-based technique using SAE features to estimate causal effects on completions; used to corroborate NLA findings.
- Dot product of summed decoder vector with unembedding vector, following Hanna et al.
- Gradient-based method to estimate the effect of zeroing a feature on a specific logit difference.
- Attribution share by system prompt paragraph relative to neutral baseline.