LLMs show statistical preemption for 120 verb-construction pairs, with surprisal driven by competing-form frequency rather than verb frequency, scaling as a power law with size, and causally shifted by controlled fine-tuning.
Ivanova and Idan Asher Blank and Nancy Kanwisher and Joshua B
11 Pith papers cite this work, alongside 91 external citations. Polarity classification is still indexing.
11
Pith papers citing it
91
external citations · external index
citation-role summary
background 2
citation-polarity summary
roles
background 2representative citing papers
Promptbreeder evolves both task prompts and the mutation prompts that improve them using LLMs, outperforming Chain-of-Thought and Plan-and-Solve on arithmetic and commonsense reasoning benchmarks.
LLM+P lets LLMs solve planning problems optimally by converting them to PDDL for classical planners and back to natural language.
Transformers on impossible-language variants show gradual grammatical sensitivity loss but sharp long-sentence generation failures, supporting generative deficiency as a link to non-attestation.
A closed-loop multi-agent system with AutoGen achieves over 98% accuracy in automated concrete barrier design and shows an 8B model outperforming 631B models.
RMCT matches the rate of target behaviors like bias-following across input perturbations to reduce sycophancy in LLMs while preserving verbalization of bias cues.
Iterated learning theory predicts and LLM experiments confirm non-monotonic compositionality during self-training, reframing model collapse as cultural transmission with matching human regularization patterns.
Sparse autoencoders applied to GPT-2 and Llama models recover semantic features accounting for 94% of peak brain encoding performance and map onto distinct cortical semantic regions across three languages.
Injecting 1% synthetic data targeting specific constructions during pre-training of GPT-2 Small boosts performance on 8 of 9 weakest BLiMP paradigms (e.g., only_npi_scope from 20.9% to 69.4%), while aggregate performance holds or improves, with one resistant case.
Supervised learning across AI systems vindicates a uniform error-driven associationism for cognition, though operating inside advanced computational structures beyond classical associationist models.
Lil-Bevo applies music pretraining, curriculum learning on sequence length, and targeted masking to small LMs in the BabyLM challenge, finding modest gains from short sequences but overall limited performance.
citing papers explorer
-
Do Language Models Know What Not to Say? Causal Evidence for Statistical Preemption in LLMs
LLMs show statistical preemption for 120 verb-construction pairs, with surprisal driven by competing-form frequency rather than verb frequency, scaling as a power law with size, and causally shifted by controlled fine-tuning.
-
Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution
Promptbreeder evolves both task prompts and the mutation prompts that improve them using LLMs, outperforming Chain-of-Thought and Plan-and-Solve on arithmetic and commonsense reasoning benchmarks.
-
LLM+P: Empowering Large Language Models with Optimal Planning Proficiency
LLM+P lets LLMs solve planning problems optimally by converting them to PDDL for classical planners and back to natural language.
-
When transformers learn "impossible" languages, what do they learn?
Transformers on impossible-language variants show gradual grammatical sensitivity loss but sharp long-sentence generation failures, supporting generative deficiency as a link to non-attestation.
-
A Lightweight Multi-Agent Framework for Automated Concrete Barrier Design
A closed-loop multi-agent system with AutoGen achieves over 98% accuracy in automated concrete barrier design and shows an 8B model outperforming 631B models.
-
Consistency Training while Mitigating Obfuscation via Rate Matching
RMCT matches the rate of target behaviors like bias-following across input perturbations to reduce sycophancy in LLMs while preserving verbalization of bias cues.
-
Model Collapse as Cultural Evolution
Iterated learning theory predicts and LLM experiments confirm non-monotonic compositionality during self-training, reframing model collapse as cultural transmission with matching human regularization patterns.
-
Sparse Autoencoders Map Brain-LLM Alignment onto Cortical Semantic Topography
Sparse autoencoders applied to GPT-2 and Llama models recover semantic features accounting for 94% of peak brain encoding performance and map onto distinct cortical semantic regions across three languages.
-
Heterogeneity in Formal Linguistic Competence of Language Models: Is Data the Real Bottleneck?
Injecting 1% synthetic data targeting specific constructions during pre-training of GPT-2 Small boosts performance on 8 of 9 weakest BLiMP paradigms (e.g., only_npi_scope from 20.9% to 69.4%), while aggregate performance holds or improves, with one resistant case.
-
The New Associationism: Lessons from Deep Learning
Supervised learning across AI systems vindicates a uniform error-driven associationism for cognition, though operating inside advanced computational structures beyond classical associationist models.
-
Lil-Bevo: Explorations of Strategies for Training Language Models in More Humanlike Ways
Lil-Bevo applies music pretraining, curriculum learning on sequence length, and targeted masking to small LMs in the BabyLM challenge, finding modest gains from short sequences but overall limited performance.