# HOMURA: Taming the Sand-Glass for Time-Constrained LLM Translation via Reinforcement Learning

arXiv: https://arxiv.org/abs/2601.10187 · HTML: https://mlsanity.com/paper/2601.10187/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 27 of its 45 references that are in the map.
- First posted on arXiv 15 Jan 2026.

## Builds on (most cited first, 27 of 27)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models](https://mlsanity.com/paper/2506.05176/) (2025)
- [How Good Are GPT Models at Machine Translation? A Comprehensive Evaluation](https://mlsanity.com/paper/2302.09210/) (2023)
- [A Long Way to Go: Investigating Length Correlations in RLHF](https://mlsanity.com/paper/2310.03716/) (2023)
- [Is ChatGPT A Good Translator? Yes With GPT-4 As The Engine](https://mlsanity.com/paper/2301.08745/) (2023)
- [Multilingual Machine Translation with Large Language Models: Empirical Results and Analysis](https://mlsanity.com/paper/2304.04675/) (2023)
- [Dual Learning for Machine Translation](https://mlsanity.com/paper/1611.00179/) (2016)
- [Generative Reward Models](https://mlsanity.com/paper/2410.12832/) (2024)
- [Verbosity Bias in Preference Labeling by Large Language Models](https://mlsanity.com/paper/2310.10076/) (2023)
- [LLM-Rubric: A Multidimensional, Calibrated Approach to Automated Evaluation of Natural Language Texts](https://mlsanity.com/paper/2501.00274/) (2024)
- [WMT24++: Expanding the Language Coverage of WMT24 to 55 Languages & Dialects](https://mlsanity.com/paper/2502.12404/) (2025)
- [On The Evaluation of Machine Translation Systems Trained With Back-Translation](https://mlsanity.com/paper/1908.05204/) (2019)
- [A surprisal--duration trade-off across and within the world's languages](https://mlsanity.com/paper/2109.15000/) (2021)
- [Is 42 the Answer to Everything in Subtitling-oriented Speech Translation?](https://mlsanity.com/paper/2006.01080/) (2020)
- [VideoDubber: Machine Translation with Speech-Aware Length Control for Video Dubbing](https://mlsanity.com/paper/2211.16934/) (2022)
- [MuST-Cinema: a Speech-to-Subtitles corpus](https://mlsanity.com/paper/2002.10829/) (2020)
- [Reinforcement Learning with Token-level Feedback for Controllable Text Generation](https://mlsanity.com/paper/2403.11558/) (2024)
- [LIMIT: Less Is More for Instruction Tuning Across Evaluation Paradigms](https://mlsanity.com/paper/2311.13133/) (2023)
- [On the Implications of Verbose LLM Outputs: A Case Study in Translation Evaluation](https://mlsanity.com/paper/2410.00863/) (2024)
- [SABER: Switchable and Balanced Training for Efficient LLM Reasoning](https://mlsanity.com/paper/2508.10026/) (2025)
- [Songs Across Borders: Singable and Controllable Neural Lyric Translation](https://mlsanity.com/paper/2305.16816/) (2023)
- [Length Controlled Generation for Black-box LLMs](https://mlsanity.com/paper/2412.14656/) (2024)
- [Prompting LLMs: Length Control for Isometric Machine Translation](https://mlsanity.com/paper/2506.04855/) (2025)
- [Length Aware Speech Translation for Video Dubbing](https://mlsanity.com/paper/2506.00740/) (2025)
- [LyriCAR: A Difficulty-Aware Curriculum Reinforcement Learning Framework For Controllable Lyric Translation](https://mlsanity.com/paper/2510.19967/) (2025)

## Built on by (most cited first, 1 of 1)

- [Index-Translate: A Multilingual Translation Model Family -- Text, Speech, Controlled Dubbing, and Long-Document Translation](https://mlsanity.com/paper/2609.40181/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) have achieved remarkable strides in multilingual translation but are hindered by a systemic cross-lingual verbosity bias, rendering them unsuitable for strict time-constrained tasks like subtitling and dubbing. Current prompt-engineering approaches struggle to resolve this conflict between semantic fidelity and rigid temporal feasibility. To bridge this gap, we first introduce Sand-Glass, a benchmark specifically designed to evaluate translation under syllable-level duration constraints. Furthermore, we propose Homura, a reinforcement learning framework that explicitly optimizes the trade-off between semantic preservation and temporal compliance. By employing a constrained reinforcement learning objective featuring a novel dynamic syllable-ratio reward, Homura effectively "tames" the output length. Experimental results demonstrate that Homura significantly outperforms strong baselines, achieving precise length control that respects linguistic density hierarchies without compromising semantic adequacy.