Summary of Automated Multi-language to English Machine Translation Using Generative Pre-trained Transformers, by Elijah Pelofske et al.
Automated Multi-Language to English Machine Translation Using Generative Pre-Trained Transformers
by Elijah Pelofske, Vincent Urias, Lorie M. Liebrock
First submitted to arxiv on: 23 Apr 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
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 Automated language translation using machine learning has been a significant area of interest in the machine learning community. This study investigates the use of local Generative Pretrained Transformer (GPT) models for zero-shot, sentence-wise, multi-natural-language translation into English text. The authors benchmark 16 open-source GPT models from the Huggingface LLM repository, using translated TED Talk transcripts as a reference dataset. The models are run locally on single A100 Nvidia GPUs, and performance is measured by language translation accuracy using BLEU, GLEU, METEOR, and chrF metrics, as well as wall-clock time per sentence translation. The best-performing model for the BLEU metric is ReMM-v2-L2-13B, achieving a mean score of 0.152 across all tested languages. For GLEU, it is also ReMM-v2-L2-13B, with a mean score of 0.256. Llama2-chat-AYT-13B achieves the best chrF metric scores, with a mean score of 0.448, while ReMM-v2-L2-13B performs well for METEOR, with a mean score of 0.438. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine being able to translate languages instantly and accurately. This paper explores using special computer models called GPTs to translate sentences from many different languages into English without needing any extra training. The authors tested 16 different GPT models on translated speeches from TED Talks, measuring how well they did using different methods. They found that one model, ReMM-v2-L2-13B, was particularly good at translating languages correctly. This research could help people communicate more easily across language barriers. |
Keywords
» Artificial intelligence » Bleu » Gpt » Machine learning » Transformer » Translation » Zero shot
