Summary of Building Accurate Translation-tailored Llms with Language Aware Instruction Tuning, by Changtong Zan et al.
Building Accurate Translation-Tailored LLMs with Language Aware Instruction Tuning
by Changtong Zan, Liang Ding, Li Shen, Yibing Zhen, Weifeng Liu, Dacheng Tao
First submitted to arxiv on: 21 Mar 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 A new study proposes a two-stage fine-tuning algorithm to improve the instruction-following ability of Large Language Models (LLMs) on translation tasks. Specifically, the algorithm first tunes LLMs with maximum likelihood estimation loss on a translation dataset to elicit basic translation capabilities. Then, it introduces an extra unlikelihood loss to learn from instruction-conflicting samples, which helps reduce off-target translations and improve overall translation quality. The proposed method outperforms competitive baselines on IWSLT and WMT benchmarks, achieving average SacreBLEU and BLEURT score improvements of +5.7 and +16.4, respectively. This work is significant for developing accurate LLM-based translation models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new study helps computers understand instructions better when translating languages. The researchers developed a special training method to teach machines like Large Language Models (LLMs) to follow instructions more accurately. They tested this method on different languages and found that it improved translation quality by a lot, making it easier for people to communicate across languages. |
Keywords
* Artificial intelligence * Fine tuning * Likelihood * Translation
