#trasferlearning #t5 #google
This paper from Google introduces T5 model (Text-to-Text Transfer Transformer) and releases large scale C4 corpus (~750GB) . T5 is an large neural network model that is trained in a pre-train, fine-tune learning paradigm. In this framework all NLP tasks are reframed into a unified text-to-text-format where the input and output are always text strings. Fine-tuning of the model was done on various tasks like SQAD question answering, WMT translation, CNN/Daily mail Abstractive Summarisation, Sentiment analysis, etc
⏩ Abstract: Transfer learning, where a model is first pre-trained on a data-rich task before being fine-tuned on a downstream task, has emerged as a powerful technique in natural language processing (NLP). The effectiveness of transfer learning has given rise to a diversity of approaches, methodology, and practice. In this paper, we explore the landscape of transfer learning techniques for NLP by introducing a unified framework that converts all text-based language problems into a text-to-text format. Our systematic study compares pre-training objectives, architectures, unlabeled data sets, transfer approaches, and other factors on dozens of language understanding tasks. By combining the insights from our exploration with scale and our new "Colossal Clean Crawled Corpus'', we achieve state-of-the-art results on many benchmarks covering summarization, question answering, text classification, and more. To facilitate future work on transfer learning for NLP, we release our data set, pre-trained models, and code.
Please feel free to share out the content and subscribe to my channel :)
⏩ Subscribe - / @techvizthedatascienceguy
⏩ OUTLINE:
0:00 - Background and Overview
1:22 - Abstract
2:13 - Input and Output formatting for T5 model
4:20 - The Colossal Clean Crawled Corpus (C4 Corpus)
6:23 - Downstream tasks
7:07 - Diagram view Text-to-Text Framework
9:56 - Attention Mask Patterns
11:06 - Flow chart of exploration of unsupervised objectives
⏩ Paper Title: Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
⏩ Paper: arxiv.org/abs/1910.10683
⏩ Code: github.com/google-research/te...
⏩ Author: Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, Peter J. Liu
⏩ Organisation: Google
⏩ IMPORTANT LINKS
My T5 Blog - prakhartechviz.blogspot.com/2...
Google T5 Blog - ai.googleblog.com/2020/02/exp...
LongT5 Paper Summary: • LongT5: Efficient Text...
SpanBERT - • SpanBERT: Improving Pr...
PEGASUS - • PEGASUS: Pre-training ...
Research Paper Walkthroughs - • Simple Unsupervised Ke...
Data Augmentation in NLP - • Data Augmentation usin...
*********************************************
If you want to support me financially which totally optional and voluntary :) ❤️
You can consider buying me chai ( because i don't drink coffee :) ) at www.buymeacoffee.com/TechvizC...
*********************************************
⏩ KZread - / techvizthedatascienceguy
⏩ Blog - prakhartechviz.blogspot.com
⏩ LinkedIn - / prakhar21
⏩ Medium - / prakhar.mishra
⏩ GitHub - github.com/prakhar21
⏩ Twitter - / rattller
*********************************************
Please feel free to share out the content and subscribe to my channel :)
⏩ Subscribe - / @techvizthedatascienceguy
Tools I use for making videos :)
⏩ iPad - tinyurl.com/y39p6pwc
⏩ Apple Pencil - tinyurl.com/y5rk8txn
⏩ GoodNotes - tinyurl.com/y627cfsa
#techviz #datascienceguy #nlproc #researchpaper #naturallanguageprocessing #transformer #deeplearning