Top Conference Papers

Automatic Labeling of Semantic Roles

Pre³: Enabling Deterministic Pushdown Automata for Faster Structured LLM Generation

I0T: Embedding Standardization Method Towards Zero Modality Gap

Turning Trash into Treasure: Accelerating Inference of Large Language Models with Token Recycling

Towards the Law of Capacity Gap in Distilling Language Models

Revisiting Compositional Generalization Capability of Large Language Models Considering Instruction Following Ability

Rethinking the Role of Prompting Strategies in LLM Test-Time Scaling: A Perspective of Probability Theory

Past Meets Present: Creating Historical Analogy with Large Language Models

MiniLongBench: The Low-cost Long Context Understanding Benchmark for Large Language Models

Mapping 1,000+ Language Models via the Log-Likelihood Vector

Llama See, Llama Do: A Mechanistic Perspective on Contextual Entrainment and Distraction in LLMs

LATIM: Measuring Latent Token-to-Token Interactions in Mamba Models

IndicSynth: A Large-Scale Multilingual Synthetic Speech Dataset for Low-Resource Indian Languages

HATEDAY: Insights from a Global Hate Speech Dataset Representative of a Day on Twitter

HALOGEN : Fantastic LLM Hallucinations and Where to Find Them

From REAL to SYNTHETIC: Synthesizing Millions of Diversified and Complicated User Instructions with Attributed Grounding

Capability Salience Vector: Fine-grained Alignment of Loss and Capabilities for Downstream Task Scaling Law

Bridging the Language Gaps in Large Language Models with Inference-Time Cross-Lingual Intervention

Beyond N-Grams: Rethinking Evaluation Metrics and Strategies for Multilingual Abstractive Summarization

Between Circuits and Chomsky: Pre-pretraining on Formal Languages Imparts Linguistic Biases

All That Glitters is Not Novel: Plagiarism in AI Generated Research

A New Formulation of Zipf's Meaning-Frequency Law through Contextual Diversity

Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention

Language Models Resist Alignment: Evidence From Data Compression

Fairness through Difference Awareness: Measuring Desired Group Discrimination in LLMs

A Theory of Response Sampling in LLMs: Part Descriptive and Part Prescriptive

Byte Latent Transformer: Patches Scale Better Than Tokens

Toward Automatic Discovery of a Canine Phonetic Alphabet

PARME: Parallel Corpora for Low-Resourced Middle Eastern Languages