Pytorch-BERT-Classification

This is pytorch simple implementation of Pre-training of Deep Bidirectional Transformers for Language Understanding (BERT) by using awesome pytorch BERT library

Dataset

1. IMDB(Internet Movie Database) To test model, I use a dataset of 50,000 movie reviews taken from IMDb. It is divied into 'train', 'test' dataset and each data has 25,000 movie reviews and labels(positive, negetive). You can access to dataset with this link

2. Naver Movie review It is well scrapped dataset from Naver movie review(Korean). link

How to use it?

Follow the example

1 Train Model

There is a lot of options to check.

1. train_path : A File to train model
2. valid_path : A File to valid model
3. max_length : Maximum length of word to analysis (BERT model restrict this parameter under 512)
4. save_path : A Path to save result of BERT classfier model
5. bert_name : The name of pretrained BERT model. Defalut : bert-base-uncased ( More information about pytorch-BERT model can be found in this link )
6. bert_finetuning : If you want to fintune BERT model with classfier layer, set "True" for this option
7. dropout_p : Drop probability of BERT result vector before enter to classfier layer
8. boost : If you don't need to fine tune BERT, you can make model faster to preconvert tokens to BERT result vectors.
9. n_epochs : A number of epoches to train
10. lr : learning rate of classfier layer
11. lr_main : learning rate of BERT for fine tune
12. early_stop : A early_stop condition. If you don't want to use this options, put -1
13. batch_size : Batch size to train
14. gradient_accumulation_steps : BERT is very heavy model to handle large batch size with light GPU. So I implement gradient accumulation to handle samller batch size but almost same impact of using large batch size

python train.py --train_path source/train.csv --valid_path source/test.csv --batch_size 16 --gradient_accumulation_steps 4 --boost True 

Result

Result with hyper parameter settings

Dataset	BERT pretrained	BERT finetune	Max token Length	Best Epoch	train loss	valid loss	valid accuracy
IMDB	bert-base-uncased	True	256	1	0.0169	0.0129	0.9181
IMDB	bert-base-uncased	True	512	1	0.0151	0.0112	0.9292
IMDB	bert-base-uncased	False	256	10	0.0289	0.0276	0.8027
IMDB	bert-base-uncased	False	512	10	0.0269	0.0259	0.8194
Naver	bert-base-multilingual-cased	True	512	4	0.0135	0.0199	0.8743
Naver	bert-base-multilingual-uncased	True	512	4	0.0126	0.0198	0.8743
Naver	kobert	True	512	2	0.0145	0.0163	0.8961

Comment

Fintuning result is remarkable and stunning. But just using a BERT output(wihtout fintuning) and put it through a single linear layer is not enought to handle data.

Reference

My pytorch implementation is highly impressed by other works. Please check below to see other works.

1. https://github.com/huggingface/pytorch-transformers
2. https://towardsdatascience.com/bert-classifier-just-another-pytorch-model-881b3cf05784