EPISODE · Sep 27, 2025 · 22 MIN
VCRL: Variance-based Curriculum Reinforcement Learning for Large Language Models
from Daily Paper Cast · host Jingwen Liang, Gengyu Wang
🤗 Upvotes: 95 | cs.LG, cs.CL Authors: Guochao Jiang, Wenfeng Feng, Guofeng Quan, Chuzhan Hao, Yuewei Zhang, Guohua Liu, Hao Wang Title: VCRL: Variance-based Curriculum Reinforcement Learning for Large Language Models Arxiv: http://arxiv.org/abs/2509.19803v1 Abstract: Policy-based reinforcement learning currently plays an important role in improving LLMs on mathematical reasoning tasks. However, existing rollout-based reinforcement learning methods (GRPO, DAPO, GSPO, etc.) fail to explicitly consider LLMs' learning ability for samples of different difficulty levels, which is contrary to the human cognitive process of mathematical reasoning tasks from easy to difficult. Intuitively, we find that the variance of the rollout group's reward in RLVR partly reflects the difficulty of the current sample for LLMs. Samples that are too easy or too difficult have a lower variance, while samples with moderate difficulty have a higher variance. Based on this, we propose VCRL, a curriculum reinforcement learning framework that dynamically controls the difficulty of training samples based on the variance of group rewards. Experiments on five mathematical benchmarks and two models reveal the advantages of VCRL over the current LLM RL baselines.
What this episode covers
🤗 Upvotes: 95 | cs.LG, cs.CL Authors: Guochao Jiang, Wenfeng Feng, Guofeng Quan, Chuzhan Hao, Yuewei Zhang, Guohua Liu, Hao Wang Title: VCRL: Variance-based Curriculum Reinforcement Learning for Large Language Models Arxiv: http://arxiv.org/abs/2509.19803v1 Abstract: Policy-based reinforcement learning currently plays an important role in improving LLMs on mathematical reasoning tasks. However, existing rollout-based reinforcement learning methods (GRPO, DAPO, GSPO, etc.) fail to explicitly consider LLMs' learning ability for samples of different difficulty levels, which is contrary to the human cognitive process of mathematical reasoning tasks from easy to difficult. Intuitively, we find that the variance of the rollout group's reward in RLVR partly reflects the difficulty of the current sample for LLMs. Samples that are too easy or too difficult have a lower variance, while samples with moderate difficulty have a higher variance. Based on this, we propose VCRL, a curriculum reinforcement learning framework that dynamically controls the difficulty of training samples based on the variance of group rewards. Experiments on five mathematical benchmarks and two models reveal the advantages of VCRL over the current LLM RL baselines.
NOW PLAYING
VCRL: Variance-based Curriculum Reinforcement Learning for Large Language Models
No transcript for this episode yet
Similar Episodes
No similar episodes found.