Reinforcement Learning - mat-yz-e/- GitHub Wiki

Exploy: EXport and dePLOY Reinforcement Learning policies
https://github.com/rai-opensource/exploy

Modern Reinforcement Learning in Practice — From Code to Theory
https://walkinglabs.github.io/hands-on-modern-rl/en/

Finding the Time to Think in Real-Time RL
https://aneeshers.github.io/realtime-rl/

Hands-On Modern Reinforcement Learning
https://github.com/walkinglabs/hands-on-modern-rl

FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control
https://github.com/Holiday-Robot/FlashSAC

Delightful Policy Gradient
https://arxiv.org/abs/2603.14608v1

Learning Smooth Time-Varying Linear Policies with an Action Jacobian Penalty
https://arxiv.org/abs/2602.18312

Stable Deep Reinforcement Learning via Isotropic Gaussian Representations
https://arxiv.org/abs/2602.19373

Understanding and Improving Hyperbolic Deep Reinforcement Learning
https://github.com/Probabilistic-and-Interactive-ML/hyper-rl

FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control
https://github.com/Holiday-Robot/FlashSAC

KL-Regularized Reinforcement Learning is Designed to Mode Collapse
https://arxiv.org/abs/2510.20817

What Does Flow Matching Bring To TD Learning?
https://arxiv.org/abs/2603.04333

Learning Smooth Time-Varying Linear Policies with an Action Jacobian Penalty
https://arxiv.org/abs/2602.18312v1

Gradient Regularization for RLHF/RLVR in TRL
https://github.com/JohannesAck/gradientregularization_trl </br<

Spectral Bellman Method: Unifying Representation and Exploration in RL
https://arxiv.org/abs/2507.13181

SAC Flow: Sample-Efficient Reinforcement Learning of Flow-Based Policies via Velocity-Reparameterized Sequential Modeling
https://arxiv.org/abs/2509.25756

Maximum Likelihood Reinforcement Learning
https://arxiv.org/abs/2602.02710

Q-learning with Adjoint Matching
https://github.com/colinqiyangli/qam

Variance Reduction for Long-Horizon LLM-RL
https://yingru.notion.site/The-Optimal-Token-Baseline-399211a558b782cfa936014c0d42dfb8

Normalizing Flows are Capable Models for RL
https://arxiv.org/abs/2505.23527

LightZero: A Unified Benchmark for Monte Carlo Tree Search in General Sequential Decision Scenarios
https://github.com/opendilab/LightZero

1000 Layer Networks for Self-Supervised RL
https://wang-kevin3290.github.io/scaling-crl

Reinforcement learning in differentiable multiphysics simulation with NVIDIA Warp.
https://github.com/rewarped/rewarped

Parallelised Q-Network for Continuous Action Spaces
https://mttga.github.io/posts/pqn_continuous/

Training Critics via Flow-Matching for Scaling Compute In Value-Based RL
https://github.com/CMU-AIRe/floq

Value Flows
https://arxiv.org/abs/2510.07650

OpenEnv: Agentic Execution Environments
https://github.com/meta-pytorch/OpenEnv

Dual Goal Representations
https://arxiv.org/abs/2510.06714

Democratizing Reinforcement Learning for LLMs
https://github.com/rllm-org/rllm

floq: Training Critics via Flow-Matching for Scaling Compute in Value-Based RL
https://arxiv.org/abs/2509.06863

Flow Matching Policy Gradients
https://arxiv.org/abs/2507.21053

ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models
https://arxiv.org/abs/2505.24864

Tianshouis a reinforcement learning library based on pure PyTorch and Gymnasium
https://github.com/thu-ml/tianshou