Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.00059 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910083519086592 |
|---|---|
| author | Zhang, Zizheng Liao, Yuyang Chen, Chen He, Jian Wu, Dun Yu, Qianjin Gao, Yanqin Yang, Jin Zhang, Kailai Chng, Eng Siong Zhong, Xionghu |
| author_facet | Zhang, Zizheng Liao, Yuyang Chen, Chen He, Jian Wu, Dun Yu, Qianjin Gao, Yanqin Yang, Jin Zhang, Kailai Chng, Eng Siong Zhong, Xionghu |
| contents | Iterative code generation with Large Language Models (LLMs) can be viewed as an optimization process guided by textual feedback. However, existing LLM self-correction methods predominantly operate in a stateless, trial-and-error manner akin to first-order search, failing to leverage past problem-solving experiences. To bridge this gap, we introduce TextBFGS, a Case-Based Reasoning (CBR) framework inspired by the Quasi-Newton optimization method. Instead of retrieving raw, unstructured textual instances, TextBFGS maintains a dynamic Case Base of historical "Error-to-Operator" correction trajectories to approximate the semantic curvature (inverse Hessian matrix) of the task. Specifically, given a textual error feedback (the target problem), TextBFGS retrieves analogous historical correction patterns (Retrieve) and applies these abstract operators to refine the current code (Reuse/Revise). Furthermore, successful adaptations are continuously retained back into the Case Base (Retain), enabling a self-evolving system. Empirical evaluations on Python code optimization tasks (HumanEval, MBPP) demonstrate that TextBFGS significantly outperforms stateless baselines. It achieves superior pass rates with fewer model calls, establishing an efficient, experience-driven paradigm for LLM-based code optimization. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_00059 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TextBFGS: A Case-Based Reasoning Approach to Code Optimization via Error-Operator Retrieval Zhang, Zizheng Liao, Yuyang Chen, Chen He, Jian Wu, Dun Yu, Qianjin Gao, Yanqin Yang, Jin Zhang, Kailai Chng, Eng Siong Zhong, Xionghu Machine Learning Artificial Intelligence Iterative code generation with Large Language Models (LLMs) can be viewed as an optimization process guided by textual feedback. However, existing LLM self-correction methods predominantly operate in a stateless, trial-and-error manner akin to first-order search, failing to leverage past problem-solving experiences. To bridge this gap, we introduce TextBFGS, a Case-Based Reasoning (CBR) framework inspired by the Quasi-Newton optimization method. Instead of retrieving raw, unstructured textual instances, TextBFGS maintains a dynamic Case Base of historical "Error-to-Operator" correction trajectories to approximate the semantic curvature (inverse Hessian matrix) of the task. Specifically, given a textual error feedback (the target problem), TextBFGS retrieves analogous historical correction patterns (Retrieve) and applies these abstract operators to refine the current code (Reuse/Revise). Furthermore, successful adaptations are continuously retained back into the Case Base (Retain), enabling a self-evolving system. Empirical evaluations on Python code optimization tasks (HumanEval, MBPP) demonstrate that TextBFGS significantly outperforms stateless baselines. It achieves superior pass rates with fewer model calls, establishing an efficient, experience-driven paradigm for LLM-based code optimization. |
| title | TextBFGS: A Case-Based Reasoning Approach to Code Optimization via Error-Operator Retrieval |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2602.00059 |