TRANS-ZERO: Self-Play Incentivizes Large Language Models for Multilingual Translation Without Parallel Data
Year: 2025
Type: conference-paper
Abstract: The rise of Large Language Models (LLMs) has reshaped machine translation (MT), but multilingual MT still relies heavily on parallel data for supervised fine-tuning (SFT), facing challenges like data scarcity for lowresource languages and catastrophic forgetting.To address these issues, we propose TRANS-ZERO, a self-play framework that leverages only monolingual data and the intrinsic multilingual knowledge of LLM.TRANS-ZERO combines a novel Monte-Carlo Tree Search, G-MCTS, with preference optimization, achieving strong translation performance that rivals supervised methods.Experiments demonstrate that this approach not only matches the performance of models trained on large-scale parallel data but also excels in non-English translation directions.Further analysis reveals that G-MCTS itself significantly enhances translation quality by exploring semantically consistent candidates through iterative translations, providing a robust foundation for the framework's success.Codes are available at https: //github.com/NJUNLP/trans0
Institution: Nanjing University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4412888332