# large language models

#reinforcement learning#large language models#prompt selection#kalman filter

Kalman Meets Curriculum: Efficient Dynamic Prompt Selection for Adaptive RL Finetuning

Haodong Zhu, Yangyang Ren, Yanjing Li +4

The paper introduces Kalman-Guided Prompt Selection (KGPS), a method that treats prompt difficulty as a dynamic state estimated with a Kalman filter to adaptively choose prompts du…

cs.IR2026

Hierarchical Latent Reasoning for LLM-based Recommendation

Peiyu Hu, Siying Gu, Weihai Lu +8

The paper introduces HiLaR, a framework that uses hierarchical latent reasoning and layer-aware reinforcement optimization to improve recommendation performance of large language m…

#recommendation systems#large language models#latent reasoning#hierarchical representation

#terminal task synthesis#agent training#container environments#large language models

Meta-Task: Turning Terminal Task Synthesis into a Terminal Task for Scalable Agent Training

Zhihong Pan, Jiyuan He, Kai Zhang +5

The paper introduces Meta-Task, a framework that generates and verifies terminal tasks inside real container environments, enabling scalable training of terminal agents with high‑q…

#reinforcement learning#self-distillation#contrastive learning#large language models

Contrastive Reinforced Policy Optimization via Privileged Self-Distillation

Xingjian Wu, Junlin Liu, Xingchen Liu +6

The paper introduces Contrastive Reinforced Policy Optimization (CRPO), a method that frames on‑policy self‑distillation for large language models as a contrastive learning problem…

#data engineering#large language models#autonomous agents#benchmarking

DataClawEval: A Benchmark for Data Engineering Agents in Real Industrial Harness

Debin Meng, Jiaming Yang, Zefang Zong +4

The paper introduces DataClawEval, a benchmark that tests autonomous LLM agents on end-to-end data engineering tasks across multiple production-grade SQL and Spark engines using de…

#automated evaluation#large language models#pairwise comparison#elo rating

(Towards) Scalable Reliable Automated Evaluation with Large Language Models

Bertil Braun, Martin Forell

The paper presents a scalable framework for automatically evaluating large language model outputs using pairwise comparisons and an Elo rating system, achieving rankings that align…

#world englishes#language ideology#large language models#standardization

AI systems and the reproduction of (standard) language ideologies in World Englishes

Kingsley Ugwuanyi

The paper investigates how large language models and related AI systems reproduce standard language ideologies that privilege Inner Circle English, marginalizing non‑dominant varie…

#large language models#dynamic pruning#token-level inference#model efficiency

WIDE: Boosting Adaptive LLM Inference via Token-level Dynamic Width Pruning

Haozhe Hu, Hao Wu, Peiran Yin +3

WIDE introduces a token-level dynamic width pruning framework for large language model inference, allowing each token to selectively activate attention heads and feed‑forward chann…

#alert triage#large language models#chain-of-thought reasoning#self-training

Cybersecurity Detection Classification with Reasoning-enabled Language Models

Amol Khanna, Manu Nandan, Cristian Viorel Popa +10

The paper introduces a chain-of-thought reasoning classifier built on large language models to triage Windows endpoint security alerts, using a calibrated confidence estimator to i…

#agentic reinforcement learning#self-distillation#verifiable rewards#large language models

Group-Reflective Self-Distillation for Agentic Reinforcement Learning

Binbin Zheng, Zijun Xie, Guanqun Zhao +4

The paper introduces Group-Reflective Self-Distillation (GRSD), a method that uses a policy's own verified rollouts to generate privileged guidance for better credit assignment in…

#large language models#bias evaluation#sentiment analysis#psychology methods

From Minds to Models: The Intersection of Psychology and LLM Behaviours

Oliver Guidetti, Reza Ryan

The paper applies psychological methods, specifically a prompt-based Implicit Association Test, to examine whether ChatGPT exhibits sentiment differences across racial conditions,…

#memory-augmented agents#large language models#experience reconstruction#reinforcement learning

MemHarness: Memory Is Reconstructed, Not Replayed

Rong Wu, Daocheng Fu, Licheng Wen +10

The paper introduces MemHarness, a framework that lets large language model agents reconstruct and adapt retrieved past experiences to the current context instead of replaying them…

#kv cache management#large language models#counter-causal attention#inference optimization

Back from the Future: Key-Value Cache Management by Counter-Causal Surprise

Stephen Gould, Anton van den Hengel

The paper introduces a KV cache eviction method for large language models that removes entries predicted well by future tokens using a counter‑causal attention pass, improving memo…

#large language models#reinforcement learning#high-performance computing#reward optimization

HARGO: Heterogeneity-Aware Reward-Guided Optimization for RL Post-Training of LLMs on HPC Tasks

Tiangang Li, Xiangbo Tian

The paper introduces HARGO, a reinforcement‑learning post‑training method that weights responses by confidence and reward contrast to better align large language models with divers…

cs.CV2026

Drawing-Recode: Annotation Grounding for Parametric CAD Code Generation from Raster 2D CAD Drawings

Mingi Kim, Yongjun Kim, Hyungki Kim

Drawing-Recode is a system that converts raster 2D CAD drawings into parametric CAD code by extracting geometric features, recognizing textual annotations, grounding the annotation…

#parametric cad generation#raster drawing digitization#annotation grounding#cross-attention

#intertextuality detection#classical chinese literature#large language models#annotation benchmark

Beyond Similarity: Grounded Agentic Extraction and Expert-Adjudicated Evaluation of Intertextuality in Classical Chinese Histories

Zhaoji Wang, Wanyu Si, Jun Wang

The paper presents a method for fine‑grained intertextuality extraction between classical Chinese histories using large language models that ground reuse spans and label them with…

#mathematical reasoning#large language models#computer algebra systems#proof automation

Albilich: Steerable Proof-State Orchestration for LLM-Based Mathematical Research with CAS Integration

Ting Gong, Michael Ruofan Zeng, Yong Yang

Albilich is an open‑source agentic framework that lets large language models conduct long‑horizon mathematical research by integrating computer algebra systems, literature retrieva…

#program verification#large language models#skill self-evolution#automated reasoning

VeriSkill: A Self-Evolution Framework for Program Verification Skills

Changguo Jia, Tianqi Zhao, Zhiyou Xiao +2

The paper introduces VeriSkill, a framework that lets large language model agents automatically improve their program verification abilities by identifying skill gaps, extracting d…

#speculative decoding#large language models#long-context inference#kv cache sparsification

A Sparse Glimpse of the Whole: Train-Free Self-Speculative Decoding

Yuesong Liu, Yuan Zeng, Min Lyu +3

The paper proposes SparseSpec-L, a training-free self-speculative decoding method that uses a sparsified key‑value cache and an entropy‑based controller to speed up long‑context in…

cs.IR2026

From Understanding to Action: Feedback-Grounded Policy Discovery for Generative Recommendation

Zhi Chen, Minmao Wang, Xingchen Liu +8

The paper introduces a feedback‑driven framework that first extracts user intent and then discovers recommendation policies using outcome‑derived feedback, distilling this knowledg…

#generative recommendation#large language models#feedback‑driven policy discovery#intent modeling

nlin.CD2026

Chaos in reason: How chain-of-thought LLMs can look for an answer

Gregorio Jaca, KristÃ³f Benedek, JÃ¡nos TÃ¶rÃ¶k

The paper investigates large language models with nonlinear dynamics tools, revealing that their hidden state trajectories display chaotic signatures like sensitivity to initial co…

#chaos theory#large language models#nonlinear dynamics#transformer analysis

#scientific writing#peer review#large language models#revision verification

AutoSupervision: Closing the Feedback Loop in Scientific Workflows with Grounded Revision Verification

Haobo Li, Eunseo Jung, Wenxiao Zhao +8

The paper presents AutoSupervision, a system that automatically verifies whether manuscript revisions truly address reviewer comments by grounding the verification in evidence from…