Summary of Could We Have Had Better Multilingual Llms If English Was Not the Central Language?, by Ryandito Diandaru et al.
Could We Have Had Better Multilingual LLMs If English Was Not the Central Language?
by Ryandito Diandaru, Lucky Susanto, Zilu Tang, Ayu Purwarianti, Derry Wijaya
First submitted to arxiv on: 21 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
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 study examines the machine translation capabilities of Large Language Models (LLMs) and explores whether there are better central languages for LLMs beyond English. The researchers modeled a linear relationship between linguistic feature distances and machine translation scores, finding that the 7B Llama2 model yields above 10 BLEU when translating into all languages it has seen, which rarely happens for languages it has not seen. Most improvements in unseen language translation come from scaling up the model size rather than instruction tuning or increasing shot count. The study also reveals that syntactic similarity is not the only linguistic factor strongly correlated with machine translation scores, and that some languages (e.g., Swedish, Catalan) exhibit comparable correlation levels to English despite having less training data. These findings challenge the prevailing landscape of LLMs, suggesting that models centered around languages other than English could provide a more efficient foundation for multilingual applications. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Large Language Models are really good at translating languages they were trained on. But what happens when they need to translate languages they’ve never seen before? The researchers wanted to know if there’s a better language for these models to be based around, rather than just English. They looked at how well the model did at translating different languages and found that some languages that don’t have as much training data can still do just as well as English in certain situations. This is important because it could help us make more efficient multilingual applications. |
Keywords
» Artificial intelligence » Bleu » Instruction tuning » Translation
