Attribution methods localize anchoring signals in Qwen and Llama models; edge-level circuits transfer within a model but show sparse transfer from base to instruction-tuned variants.
Circuit Component Reuse Across Tasks in Transformer Language Models
3 Pith papers cite this work, alongside 3 external citations. Polarity classification is still indexing.
3
Pith papers citing it
3
external citations · external index
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
Circuit-based metrics from Vision Transformer internals provide better label-free proxies for generalization under distribution shift than existing methods like model confidence.
Activation patching provides evidence about neural network circuits when the choice of metric is aligned with the hypothesis and common interpretation errors are avoided.
citing papers explorer
-
Localizing Anchoring Pathways in Language Models
Attribution methods localize anchoring signals in Qwen and Llama models; edge-level circuits transfer within a model but show sparse transfer from base to instruction-tuned variants.
-
Inside-Out: Measuring Generalization in Vision Transformers Through Inner Workings
Circuit-based metrics from Vision Transformer internals provide better label-free proxies for generalization under distribution shift than existing methods like model confidence.
-
How to use and interpret activation patching
Activation patching provides evidence about neural network circuits when the choice of metric is aligned with the hypothesis and common interpretation errors are avoided.