Image de couverture de livre

Causal Drawbridges: Characterizing Gradient Blocking of Syntactic Islands in Transformer LMs

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Boguraev, Sasha, Mahowald, Kyle
Format:	Preprint
Publié:	2026
Sujets:	Computation and Language
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Causal Interventions Reveal Shared Structure Across English Filler-Gap Constructions
par: Boguraev, Sasha, et autres
Publié: (2025)

France or Spain or Germany or France: A Neural Account of Non-Redundant Redundant Disjunctions
par: Boguraev, Sasha, et autres
Publié: (2026)

Models Can and Should Embrace the Communicative Nature of Human-Generated Math
par: Boguraev, Sasha, et autres
Publié: (2024)

Linguistic Generalizations are not Rules: Impacts on Evaluation of LMs
par: Weissweiler, Leonie, et autres
Publié: (2025)

A suite of LMs comprehend puzzle statements as well as humans
par: Goldberg, Adele E, et autres
Publié: (2025)

You Can't Fight in Here! This is BBS!
par: Futrell, Richard, et autres
Publié: (2026)

Language Models Learn Rare Phenomena from Less Rare Phenomena: The Case of the Missing AANNs
par: Misra, Kanishka, et autres
Publié: (2024)

For Generated Text, Is NLI-Neutral Text the Best Text?
par: Mersinias, Michail, et autres
Publié: (2023)

How Linguistics Learned to Stop Worrying and Love the Language Models
par: Futrell, Richard, et autres
Publié: (2025)

Are Language Models More Like Libraries or Like Librarians? Bibliotechnism, the Novel Reference Problem, and the Attitudes of LLMs
par: Lederman, Harvey, et autres
Publié: (2024)

Emergent Introspection in AI is Content-Agnostic
par: Lederman, Harvey, et autres
Publié: (2026)

The Counterexample Game: Iterated Conceptual Analysis and Repair in Language Models
par: Drucker, Daniel, et autres
Publié: (2026)

Participle-Prepended Nominals Have Lower Entropy Than Nominals Appended After the Participle
par: Denlinger, Kristie, et autres
Publié: (2024)

Experimental Contexts Can Facilitate Robust Semantic Property Inference in Language Models, but Inconsistently
par: Misra, Kanishka, et autres
Publié: (2024)

Decrypting Cryptic Crosswords: Semantically Complex Wordplay Puzzles as a Target for NLP
par: Rozner, Josh, et autres
Publié: (2021)

Convergence and Divergence of Language Models under Different Random Seeds
par: Fehlauer, Finlay, et autres
Publié: (2025)

Language Models Fail to Introspect About Their Knowledge of Language
par: Song, Siyuan, et autres
Publié: (2025)

Constructions are Revealed in Word Distributions
par: Rozner, Joshua, et autres
Publié: (2025)

On Language Models' Sensitivity to Suspicious Coincidences
par: Padmanabhan, Sriram, et autres
Publié: (2025)

Both Direct and Indirect Evidence Contribute to Dative Alternation Preferences in Language Models
par: Yao, Qing, et autres
Publié: (2025)

Lil-Bevo: Explorations of Strategies for Training Language Models in More Humanlike Ways
par: Govindarajan, Venkata S, et autres
Publié: (2023)

Privileged Self-Access Matters for Introspection in AI
par: Song, Siyuan, et autres
Publié: (2025)

semantic-features: A User-Friendly Tool for Studying Contextual Word Embeddings in Interpretable Semantic Spaces
par: Ranganathan, Jwalanthi, et autres
Publié: (2025)

Tree Transformers are an Ineffective Model of Syntactic Constituency
par: Ginn, Michael
Publié: (2024)

Counterfactual Probing for the Influence of Affect and Specificity on Intergroup Bias
par: Govindarajan, Venkata S, et autres
Publié: (2023)

Do they mean 'us'? Interpreting Referring Expressions in Intergroup Bias
par: Govindarajan, Venkata S, et autres
Publié: (2024)

Tree-Planted Transformers: Unidirectional Transformer Language Models with Implicit Syntactic Supervision
par: Yoshida, Ryo, et autres
Publié: (2024)

Are BabyLMs Second Language Learners?
par: Edman, Lukas, et autres
Publié: (2024)

Language models align with human judgments on key grammatical constructions
par: Hu, Jennifer, et autres
Publié: (2024)

Strengthening Structural Inductive Biases by Pre-training to Perform Syntactic Transformations
par: Lindemann, Matthias, et autres
Publié: (2024)

Recite, Reconstruct, Recollect: Memorization in LMs as a Multifaceted Phenomenon
par: Prashanth, USVSN Sai, et autres
Publié: (2024)

How to Make LMs Strong Node Classifiers?
par: Xu, Zhe, et autres
Publié: (2024)

Investigating Syntactic Biases in Multilingual Transformers with RC Attachment Ambiguities in Italian and English
par: Kamerath, Michael, et autres
Publié: (2025)

Emergent Semantics Beyond Token Embeddings: Transformer LMs with Frozen Visual Unicode Representations
par: Bochkov, A.
Publié: (2025)

SAP: Syntactic Attention Pruning for Transformer-based Language Models
par: Lee, Tzu-Yun, et autres
Publié: (2025)

A Systematic Study of Compositional Syntactic Transformer Language Models
par: Zhao, Yida, et autres
Publié: (2025)

Compositional preference models for aligning LMs
par: Go, Dongyoung, et autres
Publié: (2023)

Humans and transformer LMs: Abstraction drives language learning
par: Jian, Jasper, et autres
Publié: (2026)

FactCheckmate: Preemptively Detecting and Mitigating Hallucinations in LMs
par: Alnuhait, Deema, et autres
Publié: (2024)

Beyond Pattern Recognition: Probing Mental Representations of LMs
par: Miller, Moritz, et autres
Publié: (2025)