← Back to all news
arXiv AI By Xiuyi Lou, Zicheng Xu, Yu-Neng Chuang, Hoang Anh Duy Le, Zhaozhuo Xu, Guanchu Wang, Vladimir Braverman

Right Answer, Wrong Tokens: Tail-Aware Credit Calibration for RLVR

Read the original on arXiv AI →

The paper introduces Tail‑Aware Credit Calibration (TACO), a method that adjusts the positive credit given to tokens in reinforcement learning for large language models (LLMs) by evaluating each token’s rarity relative to the policy’s expectations. TACO reduces credit for high tail‑risk tokens while still allowing them to influence learning, thereby mitigating the positive‑credit contamination problem seen in GRPO‑style methods. Experiments on four LLMs across eight reasoning benchmarks demonstrate that TACO consistently outperforms GRPO baselines and improves training stability for long‑horizon RL tasks.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

Related stories

arXiv AI
Oct 15

Selective Successor Eligibility Traces for RLVR

The paper introduces Selective Successor Eligibility Traces (S-trace), a critic‑free eligibility trace method designed to improve temporal credit assignment in Reinforcement Learning with Verifiable Rewards (RLVR). S-trace reweights successor importance to recover downstream credit signals under PPO-style clipping and selectively propagates them to historical tokens using entropy-based token selection. Experiments on Qwen3 models show that S-trace yields significant absolute gains in average pass@16 over GRPO and trace-style GRPO, while also producing shorter responses.

By Chaoli Mou, Xinning Chen, Zhan Zhuang, Qinghe Zheng, Deyuan Kong, Yu Zhang
llmsreinforcement-learning
More like this →
arXiv Machine Learning
Jul 31

ACPO: Asymmetric Credit Policy Optimization via Mode-Local Entropy Surrogate

arXiv:2607. 03126v3 Announce Type: replace Abstract: Outcome-supervised reinforcement learning scales to verifiable reasoning tasks, but trajectory-level rewards assign the same outcome signal to all sampled tokens, overlooking their unequal contributions to the reasoning process.

By Zijun Xie, Yuyang You, Yongzhi Li, Enlei Gong, Quan Chen, Yanhua Cheng, Peng Jiang, Binbin Zheng, Xiaolong Liu, Zeyu Chen, Yadong Mu
reinforcement-learningbenchmarks
More like this →
arXiv AI
Jul 7

ACPO: Adaptive Credit Policy Optimization via Fine-Grained Surrogate Entropy

arXiv:2607. 03126v1 Announce Type: cross Abstract: Reinforcement Learning (RL) has substantially improved the reasoning ability of large language models (LLMs), but sparse outcome rewards still make token-level credit assignment difficult.

By Zijun Xie, Yuyang You, Yongzhi Li, Enlei Gong, Zeyu Chen, Quan Chen, Yanhua Cheng, Peng Jiang, Yadong Mu
llmsreinforcement-learningbenchmarkssafety
More like this →