grpo

19 projetos partilham este topic do GitHub

Linhas conectam membros que estão mensuravelmente relacionados entre si. O tamanho do ponto reflete estrelas.

🧬 Membros

Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.6, DeepSeek-V4, GLM-5.1, InternLM3, Llama4,…

Agent Reinforcement Trainer: train multi-step agents for real-world tasks using GRPO. Give your agents…

Solve Visual Understanding with Reinforced VLMs

Skywork-R1V is an advanced multimodal AI model series developed by Skywork AI, specializing in…

verl-agent is an extension of veRL, designed for training LLM/VLM agents via RL. verl-agent is also the…

[ECCV 2026] MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE

The Continuous-Improvement Stack for Agents. Our environment data and evals power agent improvement and…

VisualThinker-R1-Zero

Explore the Multimodal “Aha Moment” on 2B Model

[NeurIPS 2025] AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive…

Multimodal RL training framework for diffusion & omni models

RLAnything (ICML 2026) & AutoTool (ICML 2026), DemyAgent: Open-Source RL for LLMs and Agentic Scenarios

Awesome-RL-for-Video-Generation

A curated list of papers on reinforcement learning for video generation

An Asynchronous Reinforcement Learning Engine for Omni-Modal Post-Training at Scale

Code For Adaptive-Boundary-Clipping GRPO. arxiv.org/pdf/2601.03895

Agentic RAG R1 Framework via Reinforcement Learning

LightRFT: Light, Efficient, Omni-modal & Reward-model Driven Reinforcement Fine-Tuning Framework

OpenThinkIMG is an end-to-end open-source framework that empowers LVLMs to think with images.

Unleashing the Power of VLMs in Autonomous Driving via Reinforcement Learning and Reasoning

DataClaw: Agentic Tailoring Multimodal Data from Raw Streams — coming soon (code, weights, dataset &…

🔗 Familias relacionadas

deepseek-r1 19 multimodal 70 embedding 16 llms 78 agent-framework 71 diffusion 68 deepseek 88 flow-matching 18

Medido a partir dos tópicos do GitHub compartilhados por ambos os projetos, ponderado pela raridade de cada tópico.