Summary of Hw-tsc’s Submission to the Ccmt 2024 Machine Translation Tasks, by Zhanglin Wu et al.
HW-TSC’s Submission to the CCMT 2024 Machine Translation Tasks
by Zhanglin Wu, Yuanchang Luo, Daimeng Wei, Jiawei Zheng, Bin Wei, Zongyao Li, Hengchao Shang, Jiaxin Guo, Shaojun Li, Weidong Zhang, Ning Xie, Hao Yang
First submitted to arxiv on: 23 Sep 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Computation and Language (cs.CL)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper presents Huawei’s Translation Services Center’s (HW-TSC) submission to the 20th China Conference on Machine Translation (CCMT 2024). The team participated in bilingual and multi-domain machine translation tasks. For these tasks, they employed various training strategies, including regularized dropout, bidirectional training, data diversification, forward translation, back translation, alternated training, curriculum learning, and transductive ensemble learning to train neural machine translation (NMT) models based on the deep Transformer-big architecture. To explore the potential of large language models (LLMs) in improving NMT systems, they fine-tuned llama2-13b as an Automatic post-editing (APE) model for the multi-domain task. By using these strategies, their submission achieved competitive results in the final evaluation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper’s main idea is to see if large language models can help improve machine translation results. Huawei’s Translation Services Center submitted their work to a big conference and did well! They used special training methods for neural machine translation (NMT) models, and then added even more tricks by using a big language model as an editor. It worked! |
Keywords
» Artificial intelligence » Curriculum learning » Dropout » Language model » Transformer » Translation
