# Rewrite to Translate, Translate to Reward: Reinforcement Learning for Source Rewriting in Machine Translation

arXiv: https://arxiv.org/abs/2606.08011 · HTML: https://mlsanity.com/paper/2606.08011/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in LLM preference optimization methods, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 17 of its 51 references that are in the map.
- First posted on arXiv 6 Jun 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [LLM preference optimization methods](https://mlsanity.com/alignment/llm-preference-optimization-methods/)

## Builds on (most cited first, 17 of 17)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://mlsanity.com/paper/2503.14476/) (2025)
- [FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning](https://mlsanity.com/paper/2307.08691/) (2023)
- [Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch](https://mlsanity.com/paper/2311.03099/) (2023)
- [8-bit Optimizers via Block-wise Quantization](https://mlsanity.com/paper/2110.02861/) (2021)
- [SWIFT:A Scalable lightWeight Infrastructure for Fine-Tuning](https://mlsanity.com/paper/2408.05517/) (2024)
- [xCOMET: Transparent Machine Translation Evaluation through Fine-grained Error Detection](https://mlsanity.com/paper/2310.10482/) (2023)
- [GEMBA-MQM: Detecting Translation Quality Error Spans with GPT-4](https://mlsanity.com/paper/2310.13988/) (2023)
- [MetricX-24: The Google Submission to the WMT 2024 Metrics Shared Task](https://mlsanity.com/paper/2410.03983/) (2024)
- [Mitigating Metric Bias in Minimum Bayes Risk Decoding](https://mlsanity.com/paper/2411.03524/) (2024)
- [Simplify-then-Translate: Automatic Preprocessing for Black-Box Machine Translation](https://mlsanity.com/paper/2005.11197/) (2020)
- [Automatic Input Rewriting Improves Translation with Large Language Models](https://mlsanity.com/paper/2502.16682/) (2025)
- [Learning from others' mistakes: Finetuning machine translation models with span-level error annotations](https://mlsanity.com/paper/2410.16509/) (2024)
- [Understanding Pre-Editing for Black-Box Neural Machine Translation](https://mlsanity.com/paper/2102.02955/) (2021)

## Abstract (arXiv)

Prior work has explored prompting large language models (LLMs) to rewrite source text before translation, with the goal of improving machine translation (MT) quality. However, we find that such prompt-based rewriting can degrade translation quality rather than enhance it, particularly when smaller LLMs, such as 4B-parameter models, are used. We argue that this limitation stems from the difficulty of controlling rewriting behavior through natural-language prompts alone: a rewrite is useful only if it leads to a better downstream translation, yet existing prompt-based methods do not explicitly optimize for this signal. To address this issue, we propose RLSR (Reinforcement Learning for Source Rewriting), a reinforcement learning framework that trains the rewriting model with a reward derived from the downstream translation-quality improvement produced by each rewrite. Experiments across six MT models and 16 language pairs show that our 4B RLSR-trained rewriting models significantly outperform both the no-rewriting baseline and same-scale prompt-based rewriting baselines, while remaining competitive with baselines that use a 235B LLM. Our models and code are available at: https://github.com/vlaks425/MT-RLSR