Let LLMs Break Free from Overthinking via Self-Braking Tuning
Paper
•
2505.14604
•
Published
•
23
PotentialApplication
-
-
AGENTIF: Benchmarking Instruction Following of Large Language Models in Agentic Scenarios
Paper • 2505.16944 • Published • 8 -
Training Step-Level Reasoning Verifiers with Formal Verification Tools
Paper • 2505.15960 • Published • 7 -
The Unreasonable Effectiveness of Entropy Minimization in LLM Reasoning
Paper • 2505.15134 • Published • 7 -
Reward Reasoning Model
Paper • 2505.14674 • Published • 37 -
General-Reasoner: Advancing LLM Reasoning Across All Domains
Paper • 2505.14652 • Published • 24 -
Fine-tuning Quantized Neural Networks with Zeroth-order Optimization
Paper • 2505.13430 • Published • 11 -
Two Experts Are All You Need for Steering Thinking: Reinforcing Cognitive Effort in MoE Reasoning Models Without Additional Training
Paper • 2505.14681 • Published • 10 -
The Hallucination Tax of Reinforcement Finetuning
Paper • 2505.13988 • Published • 8 -
TabSTAR: A Foundation Tabular Model With Semantically Target-Aware Representations
Paper • 2505.18125 • Published • 112 -
QwenLong-CPRS: Towards infty-LLMs with Dynamic Context Optimization
Paper • 2505.18092 • Published • 43 -
Synthetic Data RL: Task Definition Is All You Need
Paper • 2505.17063 • Published • 11 -
NOVER: Incentive Training for Language Models via Verifier-Free Reinforcement Learning
Paper • 2505.16022 • Published • 4 -
ARM: Adaptive Reasoning Model
Paper • 2505.20258 • Published • 45 -
Learning to Reason without External Rewards
Paper • 2505.19590 • Published • 29 -
Interleaved Reasoning for Large Language Models via Reinforcement Learning
Paper • 2505.19640 • Published • 15 -
Rethinking the Sampling Criteria in Reinforcement Learning for LLM Reasoning: A Competence-Difficulty Alignment Perspective
Paper • 2505.17652 • Published • 6 -
UFT: Unifying Supervised and Reinforcement Fine-Tuning
Paper • 2505.16984 • Published • 3 -
Universal Reasoner: A Single, Composable Plug-and-Play Reasoner for Frozen LLMs
Paper • 2505.19075 • Published • 21 -
Reinforcing Multi-Turn Reasoning in LLM Agents via Turn-Level Credit Assignment
Paper • 2505.11821 • Published • 14 -
Text2Grad: Reinforcement Learning from Natural Language Feedback
Paper • 2505.22338 • Published • 8 -
The Entropy Mechanism of Reinforcement Learning for Reasoning Language Models
Paper • 2505.22617 • Published • 131 -
Unleashing the Reasoning Potential of Pre-trained LLMs by Critique Fine-Tuning on One Problem
Paper • 2506.03295 • Published • 17 -
ConfQA: Answer Only If You Are Confident
Paper • 2506.07309 • Published • 10 -
ExpertLongBench: Benchmarking Language Models on Expert-Level Long-Form Generation Tasks with Structured Checklists
Paper • 2506.01241 • Published • 9 -
Prefix Grouper: Efficient GRPO Training through Shared-Prefix Forward
Paper • 2506.05433 • Published • 4 -
RuleReasoner: Reinforced Rule-based Reasoning via Domain-aware Dynamic Sampling
Paper • 2506.08672 • Published • 30 -
Reflect, Retry, Reward: Self-Improving LLMs via Reinforcement Learning
Paper • 2505.24726 • Published • 277 -
ShorterBetter: Guiding Reasoning Models to Find Optimal Inference Length for Efficient Reasoning
Paper • 2504.21370 • Published • 2 -
Skywork-Reward-V2: Scaling Preference Data Curation via Human-AI Synergy
Paper • 2507.01352 • Published • 56 -
Stabilizing Knowledge, Promoting Reasoning: Dual-Token Constraints for RLVR
Paper • 2507.15778 • Published • 21 -
Sculptor: Empowering LLMs with Cognitive Agency via Active Context Management
Paper • 2508.04664 • Published • 13 -
Can LLM-Generated Textual Explanations Enhance Model Classification Performance? An Empirical Study
Paper • 2508.09776 • Published • 3 -
Aryabhata: An exam-focused language model for JEE Math
Paper • 2508.08665 • Published • 16 -
Pass@k Training for Adaptively Balancing Exploration and Exploitation of Large Reasoning Models
Paper • 2508.10751 • Published • 28 -
A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems
Paper • 2508.07407 • Published • 98 -
Part I: Tricks or Traps? A Deep Dive into RL for LLM Reasoning
Paper • 2508.08221 • Published • 50 -
Prompt Orchestration Markup Language
Paper • 2508.13948 • Published • 48 -
If We May De-Presuppose: Robustly Verifying Claims through Presupposition-Free Question Decomposition
Paper • 2508.16838 • Published • 1 -
Breaking the Exploration Bottleneck: Rubric-Scaffolded Reinforcement Learning for General LLM Reasoning
Paper • 2508.16949 • Published • 24 -
VerlTool: Towards Holistic Agentic Reinforcement Learning with Tool Use
Paper • 2509.01055 • Published • 78 -
DynaGuard: A Dynamic Guardrail Model With User-Defined Policies
Paper • 2509.02563 • Published • 21 -
ΔL Normalization: Rethink Loss Aggregation in RLVR
Paper • 2509.07558 • Published • 7 -
zELO: ELO-inspired Training Method for Rerankers and Embedding Models
Paper • 2509.12541 • Published • 6 -
Evolving Language Models without Labels: Majority Drives Selection, Novelty Promotes Variation
Paper • 2509.15194 • Published • 33 -
EasySteer: A Unified Framework for High-Performance and Extensible LLM Steering
Paper • 2509.25175 • Published • 31 -
Scaling Generalist Data-Analytic Agents
Paper • 2509.25084 • Published • 20 -
From Harm to Help: Turning Reasoning In-Context Demos into Assets for Reasoning LMs
Paper • 2509.23196 • Published • 10 -
SCI-Verifier: Scientific Verifier with Thinking
Paper • 2509.24285 • Published • 10 -
LaSeR: Reinforcement Learning with Last-Token Self-Rewarding
Paper • 2510.14943 • Published • 40 -
LLM-guided Hierarchical Retrieval
Paper • 2510.13217 • Published • 21 -
MemMamba: Rethinking Memory Patterns in State Space Model
Paper • 2510.03279 • Published • 72 -
E^2Rank: Your Text Embedding can Also be an Effective and Efficient Listwise Reranker
Paper • 2510.22733 • Published • 32 -
Redefining Retrieval Evaluation in the Era of LLMs
Paper • 2510.21440 • Published • 9 -
NeuroAda: Activating Each Neuron's Potential for Parameter-Efficient Fine-Tuning
Paper • 2510.18940 • Published • 9