RL + Transformer = A General-Purpose Problem Solver
Paper
•
2501.14176
•
Published
•
28
RL+reason model
-
-
Towards General-Purpose Model-Free Reinforcement Learning
Paper • 2501.16142 • Published • 30 -
SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training
Paper • 2501.17161 • Published • 122 -
MaxInfoRL: Boosting exploration in reinforcement learning through information gain maximization
Paper • 2412.12098 • Published • 5 -
RLDG: Robotic Generalist Policy Distillation via Reinforcement Learning
Paper • 2412.09858 • Published • 2 -
Thoughts Are All Over the Place: On the Underthinking of o1-Like LLMs
Paper • 2501.18585 • Published • 61 -
o3-mini vs DeepSeek-R1: Which One is Safer?
Paper • 2501.18438 • Published • 24 -
s1: Simple test-time scaling
Paper • 2501.19393 • Published • 126 -
Process Reinforcement through Implicit Rewards
Paper • 2502.01456 • Published • 62 -
The Jumping Reasoning Curve? Tracking the Evolution of Reasoning Performance in GPT-[n] and o-[n] Models on Multimodal Puzzles
Paper • 2502.01081 • Published • 14 -
Improving Transformer World Models for Data-Efficient RL
Paper • 2502.01591 • Published • 9 -
Satori: Reinforcement Learning with Chain-of-Action-Thought Enhances LLM Reasoning via Autoregressive Search
Paper • 2502.02508 • Published • 23 -
Demystifying Long Chain-of-Thought Reasoning in LLMs
Paper • 2502.03373 • Published • 59 -
Boosting Multimodal Reasoning with MCTS-Automated Structured Thinking
Paper • 2502.02339 • Published • 22 -
A Probabilistic Inference Approach to Inference-Time Scaling of LLMs using Particle-Based Monte Carlo Methods
Paper • 2502.01618 • Published • 10 -
BOLT: Bootstrap Long Chain-of-Thought in Language Models without Distillation
Paper • 2502.03860 • Published • 24 -
Step Back to Leap Forward: Self-Backtracking for Boosting Reasoning of Language Models
Paper • 2502.04404 • Published • 24 -
Can 1B LLM Surpass 405B LLM? Rethinking Compute-Optimal Test-Time Scaling
Paper • 2502.06703 • Published • 154 -
ReasonFlux: Hierarchical LLM Reasoning via Scaling Thought Templates
Paper • 2502.06772 • Published • 21 -
LLMs Can Easily Learn to Reason from Demonstrations Structure, not content, is what matters!
Paper • 2502.07374 • Published • 41 -
Teaching Language Models to Critique via Reinforcement Learning
Paper • 2502.03492 • Published • 24 -
An Open Recipe: Adapting Language-Specific LLMs to a Reasoning Model in One Day via Model Merging
Paper • 2502.09056 • Published • 32 -
Logical Reasoning in Large Language Models: A Survey
Paper • 2502.09100 • Published • 23 -
The Danger of Overthinking: Examining the Reasoning-Action Dilemma in Agentic Tasks
Paper • 2502.08235 • Published • 58 -
video-SALMONN-o1: Reasoning-enhanced Audio-visual Large Language Model
Paper • 2502.11775 • Published • 9 -
Soundwave: Less is More for Speech-Text Alignment in LLMs
Paper • 2502.12900 • Published • 86 -
Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities?
Paper • 2502.12215 • Published • 16 -
Small Models Struggle to Learn from Strong Reasoners
Paper • 2502.12143 • Published • 38 -
Thinking Preference Optimization
Paper • 2502.13173 • Published • 17 -
Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learning
Paper • 2502.14768 • Published • 48 -
LightThinker: Thinking Step-by-Step Compression
Paper • 2502.15589 • Published • 29 -
The Relationship Between Reasoning and Performance in Large Language Models -- o3 (mini) Thinks Harder, Not Longer
Paper • 2502.15631 • Published • 9 -
Multimodal Inconsistency Reasoning (MMIR): A New Benchmark for Multimodal Reasoning Models
Paper • 2502.16033 • Published • 18 -
SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution
Paper • 2502.18449 • Published • 74 -
Self-rewarding correction for mathematical reasoning
Paper • 2502.19613 • Published • 84 -
MedVLM-R1: Incentivizing Medical Reasoning Capability of Vision-Language Models (VLMs) via Reinforcement Learning
Paper • 2502.19634 • Published • 63 -
FINEREASON: Evaluating and Improving LLMs' Deliberate Reasoning through Reflective Puzzle Solving
Paper • 2502.20238 • Published • 24 -
Visual-RFT: Visual Reinforcement Fine-Tuning
Paper • 2503.01785 • Published • 80 -
SoS1: O1 and R1-Like Reasoning LLMs are Sum-of-Square Solvers
Paper • 2502.20545 • Published • 22 -
Cognitive Behaviors that Enable Self-Improving Reasoners, or, Four Habits of Highly Effective STaRs
Paper • 2503.01307 • Published • 39 -
Efficient Test-Time Scaling via Self-Calibration
Paper • 2503.00031 • Published • 15 -
LADDER: Self-Improving LLMs Through Recursive Problem Decomposition
Paper • 2503.00735 • Published • 22 -
START: Self-taught Reasoner with Tools
Paper • 2503.04625 • Published • 114 -
Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching
Paper • 2503.05179 • Published • 46 -
R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Model
Paper • 2503.05132 • Published • 58 -
R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning
Paper • 2503.05592 • Published • 27 -
Learning from Failures in Multi-Attempt Reinforcement Learning
Paper • 2503.04808 • Published • 18 -
An Empirical Study on Eliciting and Improving R1-like Reasoning Models
Paper • 2503.04548 • Published • 8 -
MM-Eureka: Exploring Visual Aha Moment with Rule-based Large-scale Reinforcement Learning
Paper • 2503.07365 • Published • 62 -
Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models
Paper • 2503.06749 • Published • 30 -
LMM-R1: Empowering 3B LMMs with Strong Reasoning Abilities Through Two-Stage Rule-Based RL
Paper • 2503.07536 • Published • 87 -
GTR: Guided Thought Reinforcement Prevents Thought Collapse in RL-based VLM Agent Training
Paper • 2503.08525 • Published • 17 -
Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning
Paper • 2503.09516 • Published • 32 -
GoT: Unleashing Reasoning Capability of Multimodal Large Language Model for Visual Generation and Editing
Paper • 2503.10639 • Published • 51 -
VisualPRM: An Effective Process Reward Model for Multimodal Reasoning
Paper • 2503.10291 • Published • 37 -
Light-R1: Curriculum SFT, DPO and RL for Long COT from Scratch and Beyond
Paper • 2503.10460 • Published • 29 -
R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalization
Paper • 2503.10615 • Published • 17 -
Multimodal Chain-of-Thought Reasoning: A Comprehensive Survey
Paper • 2503.12605 • Published • 35 -
R1-VL: Learning to Reason with Multimodal Large Language Models via Step-wise Group Relative Policy Optimization
Paper • 2503.12937 • Published • 29 -
reWordBench: Benchmarking and Improving the Robustness of Reward Models with Transformed Inputs
Paper • 2503.11751 • Published • 16 -
DAPO: An Open-Source LLM Reinforcement Learning System at Scale
Paper • 2503.14476 • Published • 132 -
Temporal Consistency for LLM Reasoning Process Error Identification
Paper • 2503.14495 • Published • 10 -
Towards Self-Improving Systematic Cognition for Next-Generation Foundation MLLMs
Paper • 2503.12303 • Published • 7 -
SWEET-RL: Training Multi-Turn LLM Agents on Collaborative Reasoning Tasks
Paper • 2503.15478 • Published • 12 -
Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models
Paper • 2503.16419 • Published • 75 -
OpenVLThinker: An Early Exploration to Complex Vision-Language Reasoning via Iterative Self-Improvement
Paper • 2503.17352 • Published • 23 -
I Have Covered All the Bases Here: Interpreting Reasoning Features in Large Language Models via Sparse Autoencoders
Paper • 2503.18878 • Published • 119 -
SimpleRL-Zoo: Investigating and Taming Zero Reinforcement Learning for Open Base Models in the Wild
Paper • 2503.18892 • Published • 31 -
Vision-R1: Evolving Human-Free Alignment in Large Vision-Language Models via Vision-Guided Reinforcement Learning
Paper • 2503.18013 • Published • 19 -
Mind with Eyes: from Language Reasoning to Multimodal Reasoning
Paper • 2503.18071 • Published • 3 -
Inference-Time Scaling for Flow Models via Stochastic Generation and Rollover Budget Forcing
Paper • 2503.19385 • Published • 33 -
Think Twice: Enhancing LLM Reasoning by Scaling Multi-round Test-time Thinking
Paper • 2503.19855 • Published • 28 -
ReSearch: Learning to Reason with Search for LLMs via Reinforcement Learning
Paper • 2503.19470 • Published • 18 -
ViLBench: A Suite for Vision-Language Process Reward Modeling
Paper • 2503.20271 • Published • 7 -
Video-R1: Reinforcing Video Reasoning in MLLMs
Paper • 2503.21776 • Published • 79 -
OThink-MR1: Stimulating multimodal generalized reasoning capabilities via dynamic reinforcement learning
Paper • 2503.16081 • Published • 27 -
A Survey of Efficient Reasoning for Large Reasoning Models: Language, Multimodality, and Beyond
Paper • 2503.21614 • Published • 39 -
Exploring Data Scaling Trends and Effects in Reinforcement Learning from Human Feedback
Paper • 2503.22230 • Published • 44 -
Open-Reasoner-Zero: An Open Source Approach to Scaling Up Reinforcement Learning on the Base Model
Paper • 2503.24290 • Published • 62 -
What, How, Where, and How Well? A Survey on Test-Time Scaling in Large Language Models
Paper • 2503.24235 • Published • 54 -
Efficient Inference for Large Reasoning Models: A Survey
Paper • 2503.23077 • Published • 46 -
Exploring the Effect of Reinforcement Learning on Video Understanding: Insights from SEED-Bench-R1
Paper • 2503.24376 • Published • 38 -
Z1: Efficient Test-time Scaling with Code
Paper • 2504.00810 • Published • 26 -
Improved Visual-Spatial Reasoning via R1-Zero-Like Training
Paper • 2504.00883 • Published • 64 -
Understanding R1-Zero-Like Training: A Critical Perspective
Paper • 2503.20783 • Published • 52 -
Inference-Time Scaling for Generalist Reward Modeling
Paper • 2504.02495 • Published • 55 -
Rethinking RL Scaling for Vision Language Models: A Transparent, From-Scratch Framework and Comprehensive Evaluation Scheme
Paper • 2504.02587 • Published • 30 -
Rethinking Reflection in Pre-Training
Paper • 2504.04022 • Published • 79 -
Quantization Hurts Reasoning? An Empirical Study on Quantized Reasoning Models
Paper • 2504.04823 • Published • 31 -
VAPO: Efficient and Reliable Reinforcement Learning for Advanced Reasoning Tasks
Paper • 2504.05118 • Published • 25 -
Why Reasoning Matters? A Survey of Advancements in Multimodal Reasoning (v1)
Paper • 2504.03151 • Published • 14 -
VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning
Paper • 2504.06958 • Published • 11 -
DeepSeek-R1 Thoughtology: Let's <think> about LLM Reasoning
Paper • 2504.07128 • Published • 85 -
VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model
Paper • 2504.07615 • Published • 32 -
VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning
Paper • 2504.08837 • Published • 43 -
DUMP: Automated Distribution-Level Curriculum Learning for RL-based LLM Post-training
Paper • 2504.09710 • Published • 19 -
TinyLLaVA-Video-R1: Towards Smaller LMMs for Video Reasoning
Paper • 2504.09641 • Published • 16 -
Reasoning Models Can Be Effective Without Thinking
Paper • 2504.09858 • Published • 12 -
xVerify: Efficient Answer Verifier for Reasoning Model Evaluations
Paper • 2504.10481 • Published • 84 -
Efficient Reasoning Models: A Survey
Paper • 2504.10903 • Published • 18 -
Efficient Process Reward Model Training via Active Learning
Paper • 2504.10559 • Published • 13 -
A Minimalist Approach to LLM Reasoning: from Rejection Sampling to Reinforce
Paper • 2504.11343 • Published • 18 -
ReTool: Reinforcement Learning for Strategic Tool Use in LLMs
Paper • 2504.11536 • Published • 60 -
SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Models
Paper • 2504.11468 • Published • 29 -
NoisyRollout: Reinforcing Visual Reasoning with Data Augmentation
Paper • 2504.13055 • Published • 19 -
Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?
Paper • 2504.13837 • Published • 128 -
Learning to Reason under Off-Policy Guidance
Paper • 2504.14945 • Published • 85 -
FlowReasoner: Reinforcing Query-Level Meta-Agents
Paper • 2504.15257 • Published • 46 -
ToolRL: Reward is All Tool Learning Needs
Paper • 2504.13958 • Published • 44 -
OTC: Optimal Tool Calls via Reinforcement Learning
Paper • 2504.14870 • Published • 33 -
TTRL: Test-Time Reinforcement Learning
Paper • 2504.16084 • Published • 115 -
VisuLogic: A Benchmark for Evaluating Visual Reasoning in Multi-modal Large Language Models
Paper • 2504.15279 • Published • 75 -
Process Reward Models That Think
Paper • 2504.16828 • Published • 16 -
Skywork R1V2: Multimodal Hybrid Reinforcement Learning for Reasoning
Paper • 2504.16656 • Published • 57 -
Benchmarking Multimodal Mathematical Reasoning with Explicit Visual Dependency
Paper • 2504.18589 • Published • 11 -
Reinforcement Learning for Reasoning in Large Language Models with One Training Example
Paper • 2504.20571 • Published • 94 -
WebThinker: Empowering Large Reasoning Models with Deep Research Capability
Paper • 2504.21776 • Published • 57 -
Phi-4-Mini-Reasoning: Exploring the Limits of Small Reasoning Language Models in Math
Paper • 2504.21233 • Published • 47 -
Phi-4-reasoning Technical Report
Paper • 2504.21318 • Published • 50 -
100 Days After DeepSeek-R1: A Survey on Replication Studies and More Directions for Reasoning Language Models
Paper • 2505.00551 • Published • 37 -
AdaR1: From Long-CoT to Hybrid-CoT via Bi-Level Adaptive Reasoning Optimization
Paper • 2504.21659 • Published • 12 -
Llama-Nemotron: Efficient Reasoning Models
Paper • 2505.00949 • Published • 35 -
RM-R1: Reward Modeling as Reasoning
Paper • 2505.02387 • Published • 77 -
Optimizing Chain-of-Thought Reasoners via Gradient Variance Minimization in Rejection Sampling and RL
Paper • 2505.02391 • Published • 24 -
R1-Reward: Training Multimodal Reward Model Through Stable Reinforcement Learning
Paper • 2505.02835 • Published • 26 -
Unified Multimodal Chain-of-Thought Reward Model through Reinforcement Fine-Tuning
Paper • 2505.03318 • Published • 93 -
Absolute Zero: Reinforced Self-play Reasoning with Zero Data
Paper • 2505.03335 • Published • 175 -
ZeroSearch: Incentivize the Search Capability of LLMs without Searching
Paper • 2505.04588 • Published • 65 -
Scalable Chain of Thoughts via Elastic Reasoning
Paper • 2505.05315 • Published • 25 -
X-Reasoner: Towards Generalizable Reasoning Across Modalities and Domains
Paper • 2505.03981 • Published • 14 -
MiMo: Unlocking the Reasoning Potential of Language Model -- From Pretraining to Posttraining
Paper • 2505.07608 • Published • 80 -
DanceGRPO: Unleashing GRPO on Visual Generation
Paper • 2505.07818 • Published • 29 -
Skywork-VL Reward: An Effective Reward Model for Multimodal Understanding and Reasoning
Paper • 2505.07263 • Published • 29 -
AM-Thinking-v1: Advancing the Frontier of Reasoning at 32B Scale
Paper • 2505.08311 • Published • 16 -
Bring Reason to Vision: Understanding Perception and Reasoning through Model Merging
Paper • 2505.05464 • Published • 10 -
Omni-R1: Do You Really Need Audio to Fine-Tune Your Audio LLM?
Paper • 2505.09439 • Published • 8 -
Beyond 'Aha!': Toward Systematic Meta-Abilities Alignment in Large Reasoning Models
Paper • 2505.10554 • Published • 119 -
WorldPM: Scaling Human Preference Modeling
Paper • 2505.10527 • Published • 33 -
AdaptThink: Reasoning Models Can Learn When to Think
Paper • 2505.13417 • Published • 79 -
Thinkless: LLM Learns When to Think
Paper • 2505.13379 • Published • 50 -
VisionReasoner: Unified Visual Perception and Reasoning via Reinforcement Learning
Paper • 2505.12081 • Published • 17 -
ExTrans: Multilingual Deep Reasoning Translation via Exemplar-Enhanced Reinforcement Learning
Paper • 2505.12996 • Published • 3 -
Optimizing Anytime Reasoning via Budget Relative Policy Optimization
Paper • 2505.13438 • Published • 35 -
VisualQuality-R1: Reasoning-Induced Image Quality Assessment via Reinforcement Learning to Rank
Paper • 2505.14460 • Published • 30 -
Reward Reasoning Model
Paper • 2505.14674 • Published • 35 -
Web-Shepherd: Advancing PRMs for Reinforcing Web Agents
Paper • 2505.15277 • Published • 101 -
Tool-Star: Empowering LLM-Brained Multi-Tool Reasoner via Reinforcement Learning
Paper • 2505.16410 • Published • 56 -
Pixel Reasoner: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning
Paper • 2505.15966 • Published • 51 -
GRIT: Teaching MLLMs to Think with Images
Paper • 2505.15879 • Published • 12 -
VeriThinker: Learning to Verify Makes Reasoning Model Efficient
Paper • 2505.17941 • Published • 24 -
Synthetic Data RL: Task Definition Is All You Need
Paper • 2505.17063 • Published • 10 -
One RL to See Them All: Visual Triple Unified Reinforcement Learning
Paper • 2505.18129 • Published • 59 -
G1: Bootstrapping Perception and Reasoning Abilities of Vision-Language Model via Reinforcement Learning
Paper • 2505.13426 • Published • 13 -
Active-O3: Empowering Multimodal Large Language Models with Active Perception via GRPO
Paper • 2505.21457 • Published • 14 -
The Entropy Mechanism of Reinforcement Learning for Reasoning Language Models
Paper • 2505.22617 • Published • 122 -
Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO
Paper • 2505.22453 • Published • 45 -
Sherlock: Self-Correcting Reasoning in Vision-Language Models
Paper • 2505.22651 • Published • 50 -
Skywork Open Reasoner 1 Technical Report
Paper • 2505.22312 • Published • 53 -
Advancing Multimodal Reasoning via Reinforcement Learning with Cold Start
Paper • 2505.22334 • Published • 36 -
Universal Reasoner: A Single, Composable Plug-and-Play Reasoner for Frozen LLMs
Paper • 2505.19075 • Published • 21 -
DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning
Paper • 2505.14362 • Published • 1 -
Table-R1: Inference-Time Scaling for Table Reasoning
Paper • 2505.23621 • Published • 92 -
GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning
Paper • 2505.17022 • Published • 26 -
LLMs for Engineering: Teaching Models to Design High Powered Rockets
Paper • 2504.19394 • Published • 13 -
ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models
Paper • 2505.24864 • Published • 131 -
DINO-R1: Incentivizing Reasoning Capability in Vision Foundation Models
Paper • 2505.24025 • Published • 27 -
MoDoMoDo: Multi-Domain Data Mixtures for Multimodal LLM Reinforcement Learning
Paper • 2505.24871 • Published • 21 -
Harnessing Negative Signals: Reinforcement Distillation from Teacher Data for LLM Reasoning
Paper • 2505.24850 • Published • 9 -
Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning
Paper • 2506.01939 • Published • 164 -
SRPO: Enhancing Multimodal LLM Reasoning via Reflection-Aware Reinforcement Learning
Paper • 2506.01713 • Published • 46 -
AReaL: A Large-Scale Asynchronous Reinforcement Learning System for Language Reasoning
Paper • 2505.24298 • Published • 25 -
Reflect, Retry, Reward: Self-Improving LLMs via Reinforcement Learning
Paper • 2505.24726 • Published • 222 -
OThink-R1: Intrinsic Fast/Slow Thinking Mode Switching for Over-Reasoning Mitigation
Paper • 2506.02397 • Published • 36 -
OpenThoughts: Data Recipes for Reasoning Models
Paper • 2506.04178 • Published • 41 -
Critique-GRPO: Advancing LLM Reasoning with Natural Language and Numerical Feedback
Paper • 2506.03106 • Published • 6 -
Reinforcement Pre-Training
Paper • 2506.08007 • Published • 234 -
Confidence Is All You Need: Few-Shot RL Fine-Tuning of Language Models
Paper • 2506.06395 • Published • 123 -
Magistral
Paper • 2506.10910 • Published • 60 -
MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention
Paper • 2506.13585 • Published • 240 -
VGR: Visual Grounded Reasoning
Paper • 2506.11991 • Published • 19 -
A Technical Study into Small Reasoning Language Models
Paper • 2506.13404 • Published • 9 -
Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs
Paper • 2506.14245 • Published • 35 -
Truncated Proximal Policy Optimization
Paper • 2506.15050 • Published • 10 -
Revisiting Reinforcement Learning for LLM Reasoning from A Cross-Domain Perspective
Paper • 2506.14965 • Published • 42 -
RLPR: Extrapolating RLVR to General Domains without Verifiers
Paper • 2506.18254 • Published • 30 -
ReasonFlux-PRM: Trajectory-Aware PRMs for Long Chain-of-Thought Reasoning in LLMs
Paper • 2506.18896 • Published • 26 -
GRPO-CARE: Consistency-Aware Reinforcement Learning for Multimodal Reasoning
Paper • 2506.16141 • Published • 25 -
SRFT: A Single-Stage Method with Supervised and Reinforcement Fine-Tuning for Reasoning
Paper • 2506.19767 • Published • 12