modelscope / ms-swift
Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.8, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM5.3, Gemma4, Llava, Phi4, ...) (AAAI 2025).
modelscope / ms-swift
Orchestra-Research / AI-Research-SKILLs
Comprehensive open-source library of AI research and engineering skills for any AI model. Package the skills and your claude code/codex/gemini agent will be an AI research agent with full horsepower. Maintained by Orchestra Research.
OpenPipe / ART
Agent Reinforcement Trainer: train multi-step agents for real-world tasks using GRPO. Give your agents on-the-job training. Reinforcement learning for Qwen3.6, GPT-OSS, Llama, and more!
oumi-ai / oumi
Easily fine-tune (SFT/RL), evaluate, and deploy Qwen, Gemma, or any open agentic LLM/VLM!
rasbt / reasoning-from-scratch
Implement a reasoning LLM in PyTorch from scratch, step by step
walkinglabs / hands-on-modern-rl
🚀 An open-source, hands-on curriculum bridging the gap from basic RL concepts to LLM alignment, RLVR, and advanced Agentic systems.
datawhalechina / diy-llm
Covers pre-training data, Tokenizer, Transformer, MoE,distributed training, Scaling Laws, inference & alignment .6 progressive code assignments for full-stack LLM learning | 涵盖预训练数据、分词器、Transformer、MoE、分布式训练、缩放定律、推理与对齐，6 项渐进代码作业，掌握 LLM 全栈知识
verl-project / verl-omni
Multimodal RL training framework for diffusion & omni models
JudgmentLabs / judgeval
The Continuous-Improvement Stack for Agents. Our environment data and evals power agent improvement and monitoring.
whitecircle / halo
Halo is an open-source framework built by White Circle for training large language and multimodal models
redai-studio / Relax
An Asynchronous Reinforcement Learning Engine for Omni-Modal Post-Training at Scale
Haozhe-Xing / agent_learning
A systematic AI Agent development tutorial covering LLM agents, RAG, tool use, memory systems, multi-agent systems, LangChain, LangGraph, MCP, and agentic RL.｜从零开始学 AI Agent 开发 | 系统、全面、实战导向的 Agent 开发教程 | 每日自动追踪 arXiv 最新论文 | Learn AI Agent Development from Scratch
YYHDBL / shopping-grpo-longhorizon
面向长程购物 Agent 的可复现后训练
adithya-s-k / FineEnvs
FineEnvs — RL Environments 101: building and scaling RL environments in the age of LLMs
DaoyuanLi2816 / mini-verl
verl for a single consumer GPU. PPO, GRPO and on-policy distillation on NVIDIA GPUs.
SergioTermann / awesome-post-training-RL
A curated list of papers on reinforcement learning post-training for LLMs
michaelc-yu / cs336-llm-systems
Large language model systems implementation and optimization: Transformers, BPE tokenization, FlashAttention, Triton, distributed training, GPU profiling, and GRPO reasoning RL
just-a-go / ER-TW-GRPO
—
haha-yuki-haha / AUGUstus
🚗 AUGUstus: Autonomous Driving VLA Model with Unified Generation and Understanding. 🏋️ 512 H800 GPUs · ~2 days · 318K+ video hours · 513B+ training tokens. ⚡ 12 Hz on-vehicle deployment. SFT + GRPO + NAVSIM.
stonewst / P2T
P2T: Unlocking Token Rewards via Training-Free Reward Attribution (CVPR 2026) - official code
zsxm1998 / awesome-perception-aware-rlvr
Unified reproductions, benchmarking & evaluation for perception-aware RLVR and OPD in VLMs. 19 methods · 30+ benchmarks · Official code of CGPO (ACM MM 2026 Oral).
LGAI-Research / PCBWorld
A reinforcement learning environment for PCB autorouting, built on KiCad's push-and-shove router. Includes a decoder-only PPO/GRPO agent, LLM tool-calling and rule-based baselines, and a uniform 3-stage evaluation pipeline. The routing engine ships separately as the PCBWorld-Engine submodule (GPLv3).
waltstephen / UMM-Reflection
Learning Native Reflection in Unified Models: reflection SFT and multi-round Flow-GRPO RL for BAGEL
zhiyuxu03 / AV-GRPO
Official implementation of AV-GRPO: dataset, training, and inference.
VeRITAS-wuren / OpenCSR
An explainable and reproducible post-training system for tool-using customer service agents with SFT, GRPO, verl and vLLM.
just-a-go / SOVA
Official implementation of SOVA: Strict Outcome-Conditioned Virtual Advantages for Video Reasoning. Correctness-oriented calibration for TW-GRPO without extra rollouts.
infinitylogesh / watercolor-painting-grpo
Painting with RL , Teaching a Qwen3.5-35B model to pain with p5 js using GRPO