BERT_colab.py

# -*- coding: utf-8 -*-
"""142project.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1SoXREjCqnJG3zfYju0utrwPO7fhYv-7z
"""

!pip3 install simpletransformers

# Commented out IPython magic to ensure Python compatibility.
# %%writefile setup.sh
# 
# git clone https://github.com/NVIDIA/apex
# pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./apex

!sh setup.sh

from simpletransformers.classification import ClassificationModel
import pandas as pd
import logging
from sklearn.model_selection import train_test_split
from bs4 import BeautifulSoup

logging.basicConfig(level=logging.INFO)
transformers_logger = logging.getLogger('transformers')
transformers_logger.setLevel(logging.WARNING)

def dataForBert(data):
  reviews = list(data['review'])
  labels = list(data['sentiment'])
  clean_list = []
  for review, label in zip(reviews, labels):
    clean_review = BeautifulSoup(review).get_text()
    clean_review = re.sub('["]', '', clean_review)
    clean_list.append([clean_review, label])
  
  return pd.DataFrame(clean_list)

data = pd.read_csv('labeledTrainData.tsv', header=0, quoting=3, delimiter='\t')

train_df, test_df = train_test_split(data, test_size=0.2)
train_data = dataForBert(train_df)
test_data = dataForBert(test_df)

print(f'Train data has {len(train_data)} entries')
print(f'Test data has {len(test_data)} entries')

model_args={'reprocess_input_data': True,
            'overwrite_output_dir': True,
            'sliding_window': True,
            'num_train_epochs': 1,
            'do_lower_case': False,
            }

model = ClassificationModel('roberta', 'roberta-base', args=model_args)
model.train_model(train_data)

from sklearn.metrics import accuracy_score, classification_report


result, model_outputs, wrong_predictions = model.eval_model(test_data, acc=classification_report)

print(result['acc'])