← Back to all news
arXiv Computation and Language

CROCODIL: Cross-Model Code Editing with LLMs

CROCODIL is a post‑training framework designed to improve cross‑model code editing with large language models (LLMs). It addresses the problem that different LLMs, trained on distinct datasets, often make excessive edits when applied to code generated by another model. By combining a similarity reward that discourages large changes with an execution reward that ensures build and test success, CROCODIL encourages smaller, functionally correct edits.

Related stories

arXiv AI
Sep 4

When Models Edit Too Much: On the Fidelity of Minimal Code Edits

The paper investigates the problem of over‑editing by large language models when repairing code, showing that even state‑of‑the‑art models like GPT‑5.5 frequently rewrite more code than necessary. Using a benchmark of 400 BigCodeBench problems with controlled AST corruptions, the authors quantify excess edits and demonstrate that a simple preservation instruction can reduce unnecessary changes and improve pass rates. They further explore training strategies, finding that reinforcement learning yields the best balance between edit fidelity and performance retention, highlighting edit fidelity as a distinct, measurable dimension of code‑repair quality.

By Tongyao Zhu, Wei Hern Lim, Min-Yen Kan
llmsreinforcement-learningfine-tuning
More like this →
arXiv AI
Sep 10

Diffs vs. Whole Files: An Empirical Comparison of Iterative Edit-Based and Direct Generation for Flutter/Dart Code Models

The paper compares two output regimes for code-editing language models: direct generation, where the model outputs the entire modified file, and iterative diff-based generation, where the model emits a sequence of localized edits. Experiments on Flutter/Dart tasks show that direct generation consistently outperforms diff-based generation across metrics such as compilation success, token efficiency, and quality judgments. However, diff-based generation can be competitive for short, spatially localized edits, particularly in refactoring and error-handling tasks with few edit steps.

By Andrej Andrejev
llmsfine-tuning
More like this →
Hugging Face Trending Papers
Sep 3

When Models Edit Too Much: On the Fidelity of Minimal Code Edits

The paper investigates the problem of over‑editing by large language models (LLMs) when repairing code, showing that even state‑of‑the‑art models like GPT‑5.5 often rewrite more code than necessary. Using a benchmark of 400 BigCodeBench problems with controlled AST‑level corruptions, the authors quantify over‑editing and demonstrate that a simple preservation instruction can significantly reduce excess edits and cognitive complexity while improving Pass@1. They further explore post‑training strategies, finding that reinforcement learning yields the best balance between edit fidelity and performance retention, thereby establishing edit fidelity as a distinct, measurable dimension of code‑repair quality.

llmsreinforcement-learningfine-tuning
More like this →
arXiv Computation and Language
Sep 2

InComeS: Integrating Compression and Selection Mechanisms into LLMs for Efficient Model Editing

InComeS is a framework that improves large language models (LLMs) for model editing by compressing editing contexts into a key‑value cache associated with a special gist token. It adds cross‑attention modules to dynamically select the most relevant information from these gist pools, allowing efficient handling of multiple edits beyond the LLM’s context window. Experiments on diverse editing benchmarks show that InComeS enhances both effectiveness and efficiency compared to existing methods.

By Shuaiyi Li, Zhisong Zhang, Yang Deng, Chenlong Deng, Tianqing Fang, Hongming Zhang, Haitao Mi, Dong Yu, Wai Lam
llmsbenchmarks
More like this →
arXiv AI
Jun 8

SWE-IF: Aligning Code Evaluation with Human Preference

arXiv:2510. 07315v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have catalyzed vibe coding, where users leverage LLMs to generate and iteratively refine code through natural language interactions until it passes their vibe check.

By Ming Zhong, Xiang Zhou, Ting-Yun Chang, Qingze Wang, Nan Xu, Xiance Si, Dan Garrette, Shyam Upadhyay, Jeremiah Liu, Jiawei Han, Benoit Schillings, Jiao Sun
llms
More like this →