12 repositorios
Awesome GitHub RepositoriesCritic-Free Algorithms
Policy optimization methods that do not rely on explicit value function modeling.
Explore 12 awesome GitHub repositories matching part of an awesome list · Critic-Free Algorithms. Refine with filters or upvote what's useful.
Awesome Critic-Free Algorithms GitHub Repositories
Encuentra los mejores repositorios con IA.Buscaremos los repositorios que mejor coincidan usando IA.
- openrlhf/openrlhf
OpenRLHF/OpenRLHF
9,675Ver en GitHubOpenRLHF is a training framework and alignment library designed for reinforcement learning from human feedback across distributed GPU clusters. It provides tools for aligning large language models and multimodal vision-language models using algorithms such as PPO, GRPO, and DPO. The framework distinguishes itself through a distributed inference engine that overlaps sample rollout with training to increase throughput. It supports scaling to models exceeding 70 billion parameters via parameter sharding and handles long-context sequences through ring-attention sequence parallelism. The project
Robust reinforcement learning algorithm for human feedback alignment.
Pythonlarge-language-modelsopenai-o1proximal-policy-optimization - deepseek-ai/deepseek-math
deepseek-ai/DeepSeek-Math
3,346Ver en GitHubDeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
Policy optimization for mathematical reasoning in open models.
Python - minimax-ai/minimax-m1
MiniMax-AI/MiniMax-M1
3,159Ver en GitHubMiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model.
Scaling test-time compute using efficient attention mechanisms.
Python - bytedtsinghua-sia/dapo
BytedTsinghua-SIA/DAPO
1,831Ver en GitHubDAPO: an Open-source RL System from ByteDance Seed and Tsinghua AIR
Large-scale open-source reinforcement learning system for LLMs.
Python - sail-sg/understand-r1-zero
sail-sg/understand-r1-zero
1,214Ver en GitHubCritical analysis and implementation of reasoning-focused training.
Pythonllmr1-zeroreasoning - modalminds/mm-eureka
ModalMinds/MM-EUREKA
MM-EUREKA: Exploring the Frontiers of Multimodal Reasoning with Rule-based Reinforcement Learning
Stable rule-based reinforcement learning for language models.
Python - prime-rl/entropy-mechanism-of-rl
PRIME-RL/Entropy-Mechanism-of-RL
)](https://www.alphaxiv.org/abs/2505.22617)
Entropy-based mechanisms for reasoning model reinforcement learning.
Python - yaof20/flash-rl
yaof20/Flash-RL
Fast RL training with Quantized Rollouts ( Blog )
Accelerated reinforcement learning training using quantized rollouts.
Python - tsinghuac3i/unify-post-training
TsinghuaC3I/Unify-Post-Training
📝 Unified Policy Gradient Estimator • ✨ Hybrid Post-Training 🚀 Getting Started • 📊 Main Results • 💖 Acknowledgements • 📨 Contact • 🎈 Citation
Unified post-training frameworks for large language models.
Python - liziniu/remax
liziniu/ReMax
ReMax is a reinforcement learning method, tailored for reward maximization in RLHF.
Simple and efficient alignment method for large language models.
Python - mcgill-nlp/vineppo
McGill-NLP/VinePPO
Paper - Abstract - Updates - Quick Start - Installation - Download the datasets - Create Experiment Script - Single GPU Training (Only for Rho models) - Running the experiments - Code Structure - Initial SFT Checkpoints - Acknowledgement - Citation
Refined credit assignment for unlocking reasoning potential.
Python - yihedeng9/openvlthinker
yihedeng9/OpenVLThinker
Yihe Deng , Nanyun Peng , Kai-Wei Chang
Iterative SFT-RL cycles for complex vision-language reasoning.
Python
