SOTAVerified

Reinforcement Learning (RL)

Reinforcement Learning (RL) involves training an agent to take actions in an environment to maximize a cumulative reward signal. The agent interacts with the environment and learns by receiving feedback in the form of rewards or punishments for its actions. The goal of reinforcement learning is to find the optimal policy or decision-making strategy that maximizes the long-term reward.

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 1–50 of 15113 papers

TitleStatusHype
DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement LearningCode15
Introduction to Reinforcement LearningCode11
Gymnasium: A Standard Interface for Reinforcement Learning EnvironmentsCode11
DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language ModelCode9
VLM-R1: A Stable and Generalizable R1-style Large Vision-Language ModelCode9
SkyReels-V2: Infinite-length Film Generative ModelCode9
TTRL: Test-Time Reinforcement LearningCode7
MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning AttentionCode7
AReaL: A Large-Scale Asynchronous Reinforcement Learning System for Language ReasoningCode7
An Empirical Study on Reinforcement Learning for Reasoning-Search Interleaved LLM AgentsCode7
Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement LearningCode7
SimpleRL-Zoo: Investigating and Taming Zero Reinforcement Learning for Open Base Models in the WildCode7
Flow-GRPO: Training Flow Matching Models via Online RLCode7
RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement LearningCode7
Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement LearningCode7
EvoRL: A GPU-accelerated Framework for Evolutionary Reinforcement LearningCode7
Kimi k1.5: Scaling Reinforcement Learning with LLMsCode7
FinRL-Meta: Market Environments and Benchmarks for Data-Driven Financial Reinforcement LearningCode6
The Dormant Neuron Phenomenon in Deep Reinforcement LearningCode6
RAG-R1 : Incentivize the Search and Reasoning Capabilities of LLMs through Multi-query ParallelismCode5
ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language ModelsCode5
Enhancing Efficiency of Safe Reinforcement Learning via Sample ManipulationCode5
Process Reinforcement through Implicit RewardsCode5
Orbit: A Unified Simulation Framework for Interactive Robot Learning EnvironmentsCode5
Marco-o1: Towards Open Reasoning Models for Open-Ended SolutionsCode5
EnvPool: A Highly Parallel Reinforcement Learning Environment Execution EngineCode5
ZeroSearch: Incentivize the Search Capability of LLMs without SearchingCode5
Balance Reward and Safety Optimization for Safe Reinforcement Learning: A Perspective of Gradient ManipulationCode5
Understanding R1-Zero-Like Training: A Critical PerspectiveCode5
LongWriter-Zero: Mastering Ultra-Long Text Generation via Reinforcement LearningCode5
Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language ModelsCode5
Kimi-VL Technical ReportCode5
Humanoid-Gym: Reinforcement Learning for Humanoid Robot with Zero-Shot Sim2Real TransferCode5
HuatuoGPT-o1, Towards Medical Complex Reasoning with LLMsCode5
SoundMind: RL-Incentivized Logic Reasoning for Audio-Language ModelsCode5
Multi-Agent Reinforcement Learning for Autonomous Driving: A SurveyCode5
DanceGRPO: Unleashing GRPO on Visual GenerationCode5
Group-in-Group Policy Optimization for LLM Agent TrainingCode5
Fin-R1: A Large Language Model for Financial Reasoning through Reinforcement LearningCode4
QwenLong-L1: Towards Long-Context Large Reasoning Models with Reinforcement LearningCode4
R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement LearningCode4
Ray: A Distributed Framework for Emerging AI ApplicationsCode4
Pearl: A Production-ready Reinforcement Learning AgentCode4
RLlib Flow: Distributed Reinforcement Learning is a Dataflow ProblemCode4
Discovering faster matrix multiplication algorithms with reinforcement learningCode4
RL4CO: an Extensive Reinforcement Learning for Combinatorial Optimization BenchmarkCode4
DeXtreme: Transfer of Agile In-hand Manipulation from Simulation to RealityCode4
Mastering Diverse Domains through World ModelsCode4
Delving into RL for Image Generation with CoT: A Study on DPO vs. GRPOCode4
LMM-R1: Empowering 3B LMMs with Strong Reasoning Abilities Through Two-Stage Rule-Based RLCode4
Show:102550
← PrevPage 1 of 303Next →

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1PPGMean Normalized Performance0.76—Unverified
2PPOMean Normalized Performance0.58—Unverified