4 papers
To Reason or to Fabricate: Reasoning Without Shortcuts via Hint-Anchored Pairwise Aggregation
Jiuheng Lin, Chen Zhang, Yansong Feng
While reinforcement learning (RL) significantly enhances LLM reasoning, its efficacy is severely undermined by Pre-RL data overlap, where RL datasets overlap with pretraining or SF…
CLARity: Reasoning Consistency Alone Can Teach Reinforced Experts
Jiuheng Lin, Cong Jiang, Zirui Wu +2
Training expert LLMs in domains with scarce data is difficult, often relying on multiple-choice questions (MCQs). However, standard outcome-based reinforcement learning (RL) on MCQ…
Efficient Low-Resource Language Adaptation via Multi-Source Dynamic Logit Fusion
Chen Zhang, Jiuheng Lin, Zhiyuan Liao +1
Adapting large language models (LLMs) to low-resource languages (LRLs) is constrained by the scarcity of task data and computational resources. Although Proxy Tuning offers a logit…
Read it in Two Steps: Translating Extremely Low-Resource Languages with Code-Augmented Grammar Books
Chen Zhang, Jiuheng Lin, Xiao Liu +2
While large language models (LLMs) have shown promise in translating extremely low-resource languages using resources like dictionaries, the effectiveness of grammar books remains…