Research map / Large language models / Agents, code and math
Multilingual Code Snippets Training for Program Translation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use multilingual snippet denoising and translation pre-training, then fine-tune on programs | MuST-PT · CoST programs, multiple language pairs | BLEU: 46.18 -> 71.18; 25.19 -> 70.62; 85.46 -> 89.93 | Helped |
| Use multilingual snippet denoising and translation pre-training, then fine-tune on programs | MuST-PT · CodeXGLUE Java→C# | BLEU: 83.02 -> 87.37 | Helped |
| Use multilingual snippet denoising and translation pre-training, then fine-tune on programs | MuST-PT · CodeXGLUE Java→C# | CodeBLEU: 87.92 -> 86.82 | Hurt |
| Use multilingual snippet denoising and translation pre-training, then fine-tune on programs | MuST-PT · CodeXGLUE C#→Java | BLEU: 78.35 -> 85.25 | Helped |
| Use multilingual snippet denoising and translation pre-training, then fine-tune on programs | MuST-PT · CodeXGLUE C#→Java | CodeBLEU: 85.27 -> 86.09 | Helped |
| Add multilingual snippet translation (MuST) training (ablation) | Transformer · CoST program translation, 12 language pairs | BLEU: 31.22 -> 40.9; 44.38 -> 58.35; 47.34 -> 73.7 | Helped |
| Add multilingual snippet translation (MuST) training (ablation) | CodeBERT · CoST program translation, 12 language pairs | BLEU: 38.7 -> 55.5; 65.48 -> 81.09; 85.46 -> 90.47 | Helped |
| Add multilingual snippet translation (MuST) training (ablation) | TransCoder · CoST program translation, 12 language pairs | BLEU: 30.09 -> 87.09; 29.4 -> 27.7; 40.4 -> 16.88 | Mixed |
Closest papers by meaning
- QiMeng-MuPa: Mutual-Supervised Learning for Sequential-to-Parallel Code Translation (NeurIPS 2025 · similarity 0.87)
- Function-to-Style Guidance of LLMs for Code Translation (ICML 2025 · similarity 0.86)
- Automatically Testing Functional Properties of Code Translation Models (AAAI 2024 · similarity 0.85)
- CodeRosetta: Pushing the Boundaries of Unsupervised Code Translation for Parallel Programming (NeurIPS 2024 · similarity 0.85)
- NextCoder: Robust Adaptation of Code LMs to Diverse Code Edits (ICML 2025 · similarity 0.84)
- McEval: Massively Multilingual Code Evaluation (ICLR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.