These are the GLUE Test results after fine-tuning:
- Both BERT$_{BASE}$ and BERT$_{LARGE}$ outperform all systems on all tasks by a substantial margin, obtaining 4.5% and 7.0% respective average accuracy improvement
- BERT$_{LARGE}$ significantly outperforms BERT$_{BASE}$ across all tasks, especially those with very little training data

GLUE Results

The GLUE benchmark is a collection of diverse natural language understanding tasks.

Ohio State University, Columbus

This paper presents BERT fine-tuning results on 11 Natural Language Processing (NLP) tasks.

BERT Experiments

Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2018). Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805.

https://paperswithcode.com/method/bert

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

General Language Understanding Evaluation (GLUE)

The SQuAD v1.1 is a collection of 100k crowdsourced question/answer pairs.

Stanford Question Answering Dataset 
(SQuAD v1.1)

The SQuAD 2.0 task extends the SQuAD 1.1 problem definition by allowing for the possibility that no short answer exists in the provided paragraph, making the problem more realistic.

Stanford Question Answering Dataset 
(SQuAD v2.0) 

The SWAG dataset contains 113k sentence-pair completion examples that evaluate grounded common-sense inference.

Situations With Adversarial Generations (SWAG)


Ablation studies on BERT are attempts to understand the relative importance of the various pieces of BERT’s functionality by testing them in isolation.

Learn Before

Related

Learn After