Natural Language Processing
Research on language models, text understanding, generation, and computational linguistics.
cs.CL · 805 papersModel-Agnostic Lifelong LLM Safety via Externalized Attack-Defense Co-Evolution
EvoSafety introduces a novel framework for lifelong, model-agnostic LLM safety via externalized attack-defense co-evolution to counter adversarial prompts.
TruncProof: A Guardrail for LLM-based JSON Generation under Token-Length Constraints
TruncProof enables LLMs to generate grammatically valid JSON outputs while strictly adhering to predefined token length constraints.
RAG-Enhanced Large Language Models for Dynamic Content Expiration Prediction in Web Search
This paper introduces an LLM-based framework for dynamic content expiration prediction in web search, improving freshness and user experience.
LongMemEval-V2: Evaluating Long-Term Agent Memory Toward Experienced Colleagues
LongMemEval-V2 introduces a new benchmark to evaluate long-term agent memory for acquiring environment-specific experience in web environments.
Task-Adaptive Embedding Refinement via Test-time LLM Guidance
This paper introduces an LLM-guided query refinement method that adapts embedding models in real-time for challenging zero-shot search and classification tasks.
MEME: Multi-entity & Evolving Memory Evaluation
MEME is a new benchmark evaluating LLM agents' multi-entity and evolving memory, revealing severe limitations in dependency reasoning.
Routers Learn the Geometry of Their Experts: Geometric Coupling in Sparse Mixture-of-Experts
This paper reveals a geometric coupling between SMoE routers and experts, explaining how routers learn effective assignment geometry and proposing a coupling-based router.
KV-Fold: One-Step KV-Cache Recurrence for Long-Context Inference
KV-Fold enables stable, training-free long-context inference by treating the KV-cache as an accumulator, achieving high fidelity and memory efficiency.
Solve the Loop: Attractor Models for Language and Reasoning
Attractor Models introduce a stable, efficient fixed-point refinement method for iterative Transformers, significantly boosting performance in language and reasoning tasks.
Multi-Stream LLMs: Unblocking Language Models with Parallel Streams of Thoughts, Inputs and Outputs
Multi-Stream LLMs introduce parallel computation streams to unblock language models, enabling simultaneous reading, thinking, and acting for improved efficiency.
TextSeal: A Localized LLM Watermark for Provenance & Distillation Protection
TextSeal is a new LLM watermark using dual-key generation and multi-region localization for robust, distortion-free detection and distillation protection.
The Algorithmic Caricature: Auditing LLM-Generated Political Discourse Across Crisis Events
This paper audits LLM-generated political discourse during crises, finding it lacks population realism compared to observed online content.
ORCE: Order-Aware Alignment of Verbalized Confidence in Large Language Models
ORCE improves LLM verbalized confidence by decoupling its estimation from answer generation and using rank-based optimization for better calibration.
A Causal Language Modeling Detour Improves Encoder Continued Pretraining
A Causal Language Modeling detour during encoder continued pretraining boosts downstream performance, outperforming standard MLM, especially in biomedicine.
Geometric Factual Recall in Transformers
Transformers memorize facts geometrically, using embeddings that encode relational structure and an MLP as a relation-conditioned selector.
Predicting Disagreement with Human Raters in LLM-as-a-Judge Difficulty Assessment without Using Generation-Time Probability Signals
This paper proposes a method to predict disagreement between LLM-as-a-Judge difficulty ratings and human raters, without using generation-time probability signals.
ORBIT: Preserving Foundational Language Capabilities in GenRetrieval via Origin-Regulated Merging
ORBIT prevents catastrophic forgetting in GenRetrieval LLMs by regulating weight drift, preserving foundational language capabilities.
Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space
LLMs update beliefs in a low-dimensional conceptual space, showing in-context learning as trajectories through this space, grounded in structured representations.
Predicting Decisions of AI Agents from Limited Interaction through Text-Tabular Modeling
A new text-tabular model, using an "LLM-as-Observer," accurately predicts unfamiliar AI agent decisions in negotiation games from limited interactions.
Question Difficulty Estimation for Large Language Models via Answer Plausibility Scoring
Q-DAPS estimates LLM question difficulty by analyzing the entropy of answer plausibility scores, outperforming baselines and aligning with human judgment.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.