Publication
Llama-berry: Pairwise optimization for o1-like olympiad-level mathematical reasoning
Large Language Models, Reasoning
Cite this work View paper
Citation
Di Zhang, Jianbo Wu, Jingdi Lei, Tong Che, Jiatong Li, Tong Xie, Xiaoshui Huang, Shufei Zhang, Marco Pavone, Yuqiang Li, Wanli Ouyang, Dongzhan Zhou. (2025). Llama-Berry: Pairwise Optimization for O1-like Olympiad-Level Mathematical Reasoning. NAACL 2025.
About this work
LLaMA-Berry combines Monte Carlo Tree Search with iterative self-refinement for mathematical reasoning. A Pairwise Preference Reward Model compares candidate solutions, and Enhanced Borda Count converts those comparisons into a global ranking.
The evaluation examines search efficiency and problem-solving performance on benchmarks including GPQA, AIME24, and AMC23.