Language models employ a highly localized shared mechanism for filler-gap dependencies but no unified mechanism for NPI licensing, and activation patching generalizes better than supervised alignment search.
Generalizations across filler-gap dependencies in neural language models
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
verdicts
UNVERDICTED 3representative citing papers
Causal interventions reveal that coordination islands block filler-gap mechanisms in Transformers in a gradient way matching humans, yielding the hypothesis that 'and' encodes relational dependencies differently in extractable vs. conjunctive uses.
mGRADE uses learnable-spaced convolutions shown to be equivalent to delay embeddings plus a lightweight gated recurrent component to achieve low-memory multi-timescale sequence modeling.
citing papers explorer
-
Fine-Grained Analysis of Shared Syntactic Mechanisms in Language Models
Language models employ a highly localized shared mechanism for filler-gap dependencies but no unified mechanism for NPI licensing, and activation patching generalizes better than supervised alignment search.
-
Causal Drawbridges: Characterizing Gradient Blocking of Syntactic Islands in Transformer LMs
Causal interventions reveal that coordination islands block filler-gap mechanisms in Transformers in a gradient way matching humans, yielding the hypothesis that 'and' encodes relational dependencies differently in extractable vs. conjunctive uses.
-
mGRADE: Minimal Recurrent Gating Meets Delay Convolutions for Lightweight Sequence Modeling
mGRADE uses learnable-spaced convolutions shown to be equivalent to delay embeddings plus a lightweight gated recurrent component to achieve low-memory multi-timescale sequence modeling.