Generalized Language Models
Lilian Weng
A comprehensive overview of generalized pre-trained language models like BERT, GPT, and others that achieve strong performance across diverse NLP tasks without requiring labeled pre-training data. The article tracks the evolution of these models from 2018 through 2021, including releases of GPT-2, GPT-3, RoBERTa, T5, and other variants. This approach enables scaling language model training to substantially larger datasets compared to supervised learning methods used in computer vision.
Why it matters
[Updated on 2019-02-14: add ULMFiT and GPT-2.] [Updated on 2020-02-29: add ALBERT.] [Updated on 2020-10-25: add RoBERTa.] [Updated on 2020-12-13: add T5.] [Updated on 2020-12-30: add GPT-3.] [Updated on 2021-11-13: add XLNet, BART and ELECTRA; Also updated the Summary section.] I guess they are Elmo & Bert? (Image source: here) We have seen amazing progress in NLP in 2018. Large-scale pre-trained language modes like OpenAI GPT and BERT have achieved great performance on a variety of language tasks using generic model architectures. The idea is similar to how ImageNet classification pre-training helps many vision tasks (*). Even better than vision classification pre-training, this simple and powerful approach in NLP does not require labeled data for pre-training, allowing us to experiment with increased training scale, up to our very limit.
