GPT or BERT? Reviewing the tradeoffs of using Large Language Models versus smaller models
Deciding whether to use a Large Language Model or a smaller model? This video explores the tradeoffs between both approaches based on the latest research (May 2023) on the performance of these models. Slides and References below.
0:00 Introduction
0:14 Text Classification / Sentiment
1:07 Summarization
1:23 Reading Comprehension
2:12 Product Descriptions
3:32 Roblox/Honeycomb examples
4:42 Traditional NLP models
4:55 Text Classification in Finance
5:23 Named Entity Recognition in Healthcare
5:46 Tradeoffs between LLMs and Smaller models
6:23 List of Criteria
7:02 We need both types of models
Slides: https://docs.google.com/presentation/...
References:
The references here follow my opinionated view of where LLMs are going and are not meant to be comprehensive.
Performance of LLMs:
Is ChatGPT a Good Sentiment Analyzer? A Preliminary Study: https://arxiv.org/pdf/2304.04339v1.pdf
News Summarization and Evaluation in the Era of GPT-3: https://arxiv.org/pdf/2209.12356.pdf
Benchmarking Large Language Models for News Summarization: https://arxiv.org/pdf/2301.13848.pdf
A New Era of Creativity: Expert-in-the-loop Generative AI at Stitch Fix: https://multithreaded.stitchfix.com/b...
Where LLMs lose to smaller models:
Sentiment Spin: Attacking Financial Sentiment with GPT-3: https://papers.ssrn.com/sol3/papers.c...
Zero-shot Clinical Entity Recognition using ChatGPT: https://arxiv.org/pdf/2303.16416v2.pdf
━━━━━━━━━━━━━━━━━━━━━━━━━
★ Rajistics Social Media »
● Link Tree: https://linktr.ee/rajistics
● LinkedIn: / rajistics
━━━━━━━━━━━━━━━━━━━━━━━━━
Смотрите видео GPT or BERT? Reviewing the tradeoffs of using Large Language Models versus smaller models онлайн без регистрации, длительностью часов минут секунд в хорошем качестве. Это видео добавил пользователь Rajistics - data science, AI, and machine learning 03 Июнь 2023, не забудьте поделиться им ссылкой с друзьями и знакомыми, на нашем сайте его посмотрели 1,772 раз и оно понравилось 57 людям.