Summary of M-ped: Multi-prompt Ensemble Decoding For Large Language Models, by Jiaxin Guo et al.
M-Ped: Multi-Prompt Ensemble Decoding for Large Language Models
by Jiaxin Guo, Daimeng Wei, Yuanchang Luo, Shimin Tao, Hengchao Shang, Zongyao Li, Shaojun Li, Jinlong Yang, Zhanglin Wu, Zhiqiang Rao, Hao Yang
First submitted to arxiv on: 24 Dec 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 proposed multi-prompt ensemble decoding approach, dubbed Inner-Batch Ensemble, aims to improve the generation quality of Large Language Models (LLMs) by aggregating outcomes from multiple prompts. The method involves submitting n variations of prompts with a unique input X to LLMs in batch mode, calculating the ensemble probability for each token prediction by averaging the n probability distributions within the batch, and then utilizing this aggregated probability to generate the token. To facilitate efficient batch inference, a Left-Padding strategy is implemented to maintain uniform input lengths across the n prompts. The efficacy of Inner-Batch Ensemble is demonstrated through extensive experimentation on diverse NLP tasks, including machine translation, code generation, and text simplification, resulting in substantial improvements in BLEU scores, pass@k rates, and LENS metrics over conventional methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper improves how Large Language Models work with prompts. They make multiple versions of the same prompt and ask the model to generate answers for each one. Then, they combine the results from all the prompts to get a better answer. This helps the model be more accurate and helpful. The researchers tested this method on many different tasks like translating text, writing code, and making text simpler. They found that it worked well and improved the quality of the generated text. |
Keywords
» Artificial intelligence » Bleu » Inference » Nlp » Probability » Prompt » Token » Translation
