🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Qiying Yu and team just dropped the DAPO algorithm (decoupled clip and dynamic sampling policy optimization)! DAPO-Zero-32B, a fully open-source RL reasoning model, surpasses DeepSeek-R1-Zero-Qwen-32B, and scores 50 on AIME 2024 with 50% fewer steps. It is trained with

ByteDance Research Releases DAPO: A Fully Open-Sourced LLM Reinforcement Learning System at Scale Researchers from ByteDance, Tsinghua University, and the University of Hong Kong recently introduced DAPO (Dynamic Sampling Policy Optimization), an open-source large-scale

DAPO: An Open-Source LLM Reinforcement Learning System at Scale DAPO is a reinforcement learning algorithm for large-scale LLM training, achieving 50 points on AIME 2024 with Qwen2.5-32B. It introduces four key techniques to improve LLM reasoning and provides open-source

Lots of good nuggets here. Interestingly, they completely drop the KL divergence penalty and get good results. This mirrors what we're finding in our own experiments. Seems not to be so necessary for RLVR with GRPO. As a bonus, skipping it speeds up training significantly!

New RL Method thats better than GRPO! 🤯ByteDance Open Source released a new open source RL method that outperforms GRPO. DAPO or Decoupled Clip and Dynamic sAmpling Policy Optimization (DAPO) achieves 50 points on the AIME 2024 benchmark with 50% fewer training steps. TL;DR: 🏆 50%

To my RL heroes, DAPO!😂 Who’s got insights? Let’s make some magic happen. 🔗 dapo-sia.github.io will brown Alexander Doria Unsloth AI Axolotl Hugging Face (trl/open-r1)

DAPO: An Open-Source LLM Reinforcement Learning System at Scale From a joint ByteDance/Tsinghua team. Proposes the Decoupled Clip and Dynamic sAmpling Policy Optimization algorithm and fully open-sources a SOTA large-scale RL system. Both were used to achieve 50 points on AIME

DAPO: An Open-Source LLM Reinforcement Learning System at Scale It introduces DAPO, a fully open-source, large-scale RL system that boosts the chain-of-thought reasoning capabilities of LLMs. DAPO raises the upper clipping threshold (“Clip-Higher”) in PPO-style training,

Thank you AK for featuring our work. Excited to share valuable insights and open-source usefull systems to the community ! 🌟

A new RL algorithm! DAPO (Decoupled Clip and Dynamic Sampling Policy Optimization) from ByteDance Open Source is a fully open-source RL system, that improves training in long Chain-of-Thought (CoT) reasoning. It achieves 50 points on AIME 2024, surpassing DeepSeek-R1-Zero, using only

China's ByteDance presents VAPO Efficient and Reliable Reinforcement Learning for Advanced Reasoning Tasks present VAPO, Value-based Augmented Proximal Policy Optimization framework for reasoning models. a novel framework tailored for reasoning models within the value-based

🚀 Introducing VAPO (Value-based augmented PPO), our latest RL method for reasoning models. Trained from Qwen-32B-base model, VAPO achieves 60.4 on AIME 2024, outperforming DeepSeek-zero-32B and DAPO-32B📈. Built with verl project, and yes, we will open source it soon. Key

DAPO's clip high (and other insights) really works well. It has been independently reproduced! Congrats Qiying Yu !

Overlong filtering has been shown to significantly stabilize learning and improve performance. You can now use it in TRL! It simply consists in masking the loss of truncated samples. Principle proposed by Qiying Yu in DAPO, implemented by Shirin Yamani 👏

#ICLR2025 I am going to present VAPO & DAPO twice at ICLR, two SOTA LLM RL algorithms. 1. The 1-2 pm verl Expo Talk, Apr 26, Peridot 202-203 2. The 3:00-3:30 pm break, Apr 24, at the ByteDance Booth Welcome and see you there!