forestliutc/UloRL

An Ultra-Long Output Reinforcement Learning Approach for Advancing Large Language Models' Reasoning Abilities

Paper | GitHub

Overview

Recent advances in large language models (LLMs) have highlighted the potential of reinforcement learning with verifiable rewards (RLVR) to enhance reasoning capabilities through extended output sequences. However, traditional RL frameworks face inefficiencies when handling ultra-long outputs due to long-tail sequence distributions and entropy collapse during training. To address these challenges, we propose an Ultra-Long Output Reinforcement Learning (UloRL) approach for advancing large language models' reasoning abilities. Specifically, we divide ultra long output decoding into short segments, enabling efficient training by mitigating delays caused by long-tail samples. Additionally, we introduce dynamic masking of well-Mastered Positive Tokens (MPTs) to prevent entropy collapse. Experimental results demonstrate the effectiveness of our approach. On the Qwen3-30B-A3B model, RL with segment rollout achieved 2.06x increase in training speed, while RL training with 128k-token outputs improves the model's performance on AIME2025 from 70.9% to 85.1% and on BeyondAIME from 50.7% to 61.9%, even surpassing Qwen3-235B-A22B with remarkable gains.

Inference Parameters

128k setting:

temperature=0.85
top_p=0.95
top_k=20
max_tokens=131072

140k setting (with Yarn)

temperature=0.85
top_p=0.95
top_k=20
max_tokens=143360
rope_scaling: {
    "rope_type": "yarn",
    "factor": 1.5,
    "original_max_position_embeddings": 95232
  }

Results