Diff Diffusion is a diffusion-inspired large language model framework designed for efficient code editing. It predicts edit paths from an original snippet to its revised form by representing denoising as structured edit operations, and uses multi-stage training and efficient parallel decoding to focus computation on changed regions. The approach matches or outperforms autoregressive baselines while achieving up to 2.78× faster inference and a 39.0% reduction in median model latency in production.
By Xiaolong Xu, Hao Jiang, Bo Jiang, Zhiwen Deng, Ningyuan Sun, Bingzhou Chen, Jin Chen, Chenxing Wei
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
The paper introduces DECODE, a dataset of 53.6K real‑world in‑IDE edits of AI‑generated code from over 1,000 developers, and uses it to study when, why, and how developers modify such code. Findings show developers spend little effort editing and often abandon AI outputs, highlighting the need for LLMs to infer intent from edits. Experiments reveal that while frontier models can generate edits that align with developer intent, they tend to over‑edit and revert changes; fine‑tuning improves performance but does not fully close the gap.
By Jenny T. Liang, Mihika Bairathi, Wayne Chi, Ameet Talwalkar, Nishant Subramani, Valerie Chen
arXiv:2607. 17532v1 Announce Type: cross Abstract: Developers frequently write uninformative git commit messages such as "fix" or "update stuff", degrading the value of version-control history for code review, debugging, and onboarding.
By Md Rafid Haque, Poojan Narendrabhai Patel, Meetkumar Vijaybhai Raychura
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.
The paper introduces the Harmful Memorization Index (HMI) to measure when large language models (LLMs) fail to adapt to small semantic changes in coding tasks, producing code that is superficially similar to the original solution. It proposes an automated code‑rewriting pipeline that generates semantically different variants of tasks by altering core logic elements, and uses HMI to combine similarity metrics with performance degradation. Experiments on MBPP+ and BigCodeBench show that HMI does not simply grow with model size, is higher for certain logic‑edit types, and that supervised fine‑tuning can increase HMI while PPO offers a better accuracy–HMI trade‑off.
By Lizhe Zhang, Wentao Chen, Li Zhong, Letian Peng, Zilong Wang, Jingbo Shang
arXiv:2607. 06125v1 Announce Type: cross Abstract: Neural decompilation is increasingly studied as a code-generation problem, yet its evaluation methodology remains underdeveloped for modern languages.
By Raafat Abualazm, Ayman AboElhassan, Amr G. Wassal
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.
By Linghan Zhong, Aditya Thimmaiah, Jayanth Srinivasa, Milos Gligoric, Junyi Jessy Li
arXiv:2607. 12713v1 Announce Type: cross Abstract: Generated tokens are a direct driver of the cost, latency, and energy of generative AI (GAI) code editing.
By William Franz Lamberti
arXiv:2607. 18161v1 Announce Type: cross Abstract: Coding agents are increasingly used to accelerate code generation in many downstream tasks, such as fixing bugs, building applications, and prototyping.
By Alex Mathai, Shobini Iyer, Aleksandr Nogikh, Petros Maniatis, Franjo Ivancic, Junfeng Yang, Baishakhi Ray
The paper investigates how large language models (LLMs) handle bug fixing versus problem solving in competitive programming. Using a dataset of ~3,000 Codeforces submissions and their human fixes, the authors compare LLM-generated patches to human patches and assess whether LLMs prefer to modify buggy code or generate new solutions. Results show that LLMs often alter more lines than necessary and sometimes produce entirely new solutions, performing better when allowed to generate solutions from scratch rather than patching existing code.
By Alexandru Stefan Stoica, Traian Rebedea, Marian Cristian Mihaescu
Developers frequently write uninformative git commit messages such as "fix" or "update stuff", degrading the value of version-control history for code review, debugging, and onboarding. We present CommitLLM, a three-stage pipeline that generates concise, Conventional Commits-compliant messages from code diffs using a fine-tuned small language model.