IMDB-sentiment-classification-based-on-BERT

IMDB film review sentiment classification based on BERT’s supervised learning model. On the other hand, the model can be extended to other natural language multi-classification tasks.

Documents description

ALL_OUTPUT:4组实验运行结果。

BERT_BASE_DIR:谷歌预训练BERT模型文件。

DATA_DIR:训练集、验证集、测试集文件。

output_models:空文件夹，运行程序时存储输出文件。

Raw_Data:原始数据集以及数据预处理过程涉及到的一些数据文件。

IMDB Parameters:运行‘run.py’文件时需将该文件中的参数传入程序。

run.py:训练、验证、测试模型时运行的文件。

Parameters

–task_name=mrpc
–do_train=true
–do_eval=true
–do_predict=true
–data_dir=D:\XXXXXXX\IMDB\DATA_DIR
–vocab_file=BERT_BASE_DIR/vocab.txt
–bert_config_file=BERT_BASE_DIR/bert_config.json
–init_checkpoint=BERT_BASE_DIR/bert_model.ckpt
–max_seq_length=30
–train_batch_size=1
–learning_rate=2e-6
–num_train_epochs=5
–output_dir=output_models/

The optimization result

Accuracy:
0.922（validation set）
0.928（test set）

File distribution

image

GitHub

GitHub - Paris-an/IMDB-sentiment-classification-based-on-BERT at pythonawesome.com
IMDB film review sentiment classification based on BERT's supervised learning model. On the other hand, the model can be extended to other natural language multi-classification tasks. - GitHub ...
GitHubParis-an