# DPO-F+: Aligning Code Repair Feedback with Developers' Preferences

arXiv: https://arxiv.org/abs/2511.01043 · HTML: https://mlsanity.com/paper/2511.01043/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Nov 2025.

## Built on by (most cited first, 3 of 3)

- [VERITAS: Verifier-Guided Proof Search for Zero-Shot Formal Theorem Proving](https://mlsanity.com/paper/2606.19399/) (2026)
- [SynthFix: Adaptive Neuro-Symbolic Code Vulnerability Repair](https://mlsanity.com/paper/2604.17184/) (2026)
- [From Conversation to Contribution: Characterizing Coding Agent in Open-Source Software](https://mlsanity.com/paper/2607.05677/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) are increasingly used in software engineering tasks, especially code repair. However, developers often struggle to interpret model outputs, limiting effective human--AI teaming, where humans and AI work toward a shared objective. Prior work mainly optimizes generated code, giving less attention to natural-language feedback that supports comprehension and iterative improvement. We present textscDPO-f+, a framework that aligns code-repair feedback with the needs of different developer groups, including novices and proficient developers. It (1) defines feedback-alignment metrics across seven fixed dimensions with task-specific descriptions; (2) automatically constructs pairwise preference datasets from code-repair tasks; (3) fine-tunes models using Direct Preference Optimization (DPO) augmented with a reward model; and (4) provides an automated protocol for evaluating feedback quality. Empirically, textscDPO-f+ outperforms both the baseline and standard DPO in feedback accuracy and overall alignment. On novice programming tasks, textscDPO-f+ improves Pass@1 by 5.71 percentage points (pp) over the baseline and 3.30 pp over DPO. On textscSWE-Bench, it improves issue-resolution rate by 1.67 pp over DPO and 4.67 pp over the baseline. It also improves feedback alignment by both LLM judges and a human study with 200 developers: beginners preferred textscDPO-f+ in 71.5% of comparisons, with overall preference above chance (p=0.0057). By better aligning feedback with developer needs, textscDPO-f+ turns LLM assistance from a one-shot output into a collaborative sense-making workflow, enhancing code comprehension and human--AI teaming in software engineering.