Papers by Chengpeng Li
Chronos: Learning Temporal Dynamics of Reasoning Chains for Test-Time Scaling (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for testing time scales treat reasoning traces or tokens equally, ignoring substantial variations in trajectory quality and localized logical failures. |
| Approach: | They propose a chronological reasoning scorer that models each trajectory as a time series. |
| Outcome: | The proposed method achieves relative improvements of 34.21% over Pass@128 and 22.70% over Maj@135 on HMMT25, highlighting its effectiveness. |
Aligning Translation-Specific Understanding to General Understanding in Large Language Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Large Language models (LLMs) have remarkable abilities in understanding complex texts . however, understanding misalignment leads to LLMs mistakenly translating complex concepts . |
| Approach: | They propose a translation process that aligns the translation-specific understanding with the general understanding to improve translation quality and reduce translation literalness. |
| Outcome: | The proposed translation process improves translation quality and reduces translation literalness by -25% -51%. |
START: Self-taught Reasoner with Tools (2025.emnlp-main)
Copied to clipboard
Chengpeng Li, Mingfeng Xue, Zhenru Zhang, Jiaxi Yang, Beichen Zhang, Bowen Yu, Binyuan Hui, Junyang Lin, Xiang Wang, Dayiheng Liu
| Challenge: | Large Reasoning Models (LRMs) have demonstrated remarkable capabilities in complex reasoning through long chain-of-thought, yet they struggle with precise computations and algorithmic operations. |
| Approach: | They propose a training-free approach that activates LRMs’ latent tool-use capabilities through artificial hints and a framework that enables models to learn effective tool utilization through diverse hint patterns and rejection-based data synthesis. |
| Outcome: | Experiments show that START significantly improves state-of-the-art LRMs across challenging benchmarks, including competition-level mathematics (AMC23: 95.0%, AIME24: 75.6%) and graduate-level science questions (GPQA: 64.6%). |
Breaking Language Preference in Multilingual RAG via Language-Controllable Retrieval and Language-Agnostic Reasoning (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to improve retrieval accuracy and generation quality of large language models suffer from language preference. |
| Approach: | They propose a framework that explicitly disentangles multilingual RAG into language-controllable retrieval and language-agnostic reasoning. |
| Outcome: | Experimental results show that the proposed approach outperforms baselines across multilingual benchmarks. |
MuggleMath: Assessing the Impact of Query and Response Augmentation on Math Reasoning (2024.acl-long)
Copied to clipboard
Chengpeng Li, Zheng Yuan, Hongyi Yuan, Guanting Dong, Keming Lu, Jiancan Wu, Chuanqi Tan, Xiang Wang, Chang Zhou
| Challenge: | In math reasoning with large language models, fine-tuning data augmentation by query evolution and diverse reasoning paths is empirically verified effective. |
| Approach: | They propose to fine-tune data augmentation by query evolution and diverse reasoning paths. |
| Outcome: | The proposed model achieves new state-of-the-art on GSM8K and MATH. |
How Abilities in Large Language Models are Affected by Supervised Fine-tuning Data Composition (2024.acl-long)
Copied to clipboard
Guanting Dong, Hongyi Yuan, Keming Lu, Chengpeng Li, Mingfeng Xue, Dayiheng Liu, Wei Wang, Zheng Yuan, Chang Zhou, Jingren Zhou
| Challenge: | supervised fine-tuning (SFT) is a technique used to enhance multiple abilities in large language models. |
| Approach: | They propose to study the interplay of data composition between mathematical reasoning, code generation, and general human-aligning abilities during supervised fine-tuning. |
| Outcome: | The proposed model improves math reasoning and code generation with increasing data amount . the proposed model size and SFT strategies can be used to learn multiple skills with different scaling patterns. |
HintPilot: LLM-based Compiler Hint Synthesis for Code Optimization (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to optimize source code rely on invasive transformations that can introduce semantic errors and miss fine-grained compiler-level optimization opportunities. |
| Approach: | They propose a method that bridges LLM-based reasoning with traditional compilers by synthesizing compiler hints. |
| Outcome: | HintPilot achieves 6.88x speedup over -Ofast while preserving program correctness. |