Text-Classification-using-Category-Label-Embedding-and-Weakly-Supervised-Learning/pseudo_labels.py at main · LokeshSreenathJ/Text-Classification-using-Category-Label-Embedding-and-Weakly-Supervised-Learning · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
# -*- coding: utf-8 -*-
"""pseudo_labels.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1DBldKepiIfjJAyfavm71XrCJ8fUT8N2N
"""

!pip install transformers

from transformers import pipeline
classifier = pipeline("zero-shot-classification", model = "facebook/bart-large-mnli")
labels = ["politics","sports","business","technology"]

import pandas as pd
data = pd.read_csv("/content/drive/MyDrive/WeSTClass/News/combined_file.csv", header = None,error_bad_lines=False)
data.columns = ["actual_labels","document"]

from tqdm import tqdm
for i in tqdm(range(2000)):
  sequence= data.iloc[i,1]
  data.loc[i,"pseudo_label_bart-l-m"]=classifier(sequence,labels)["labels"][0]
data.to_csv("/content/drive/MyDrive/WeSTClass/News/combined_file_latest4.csv")

# For Movies:

from transformers import pipeline
classifier = pipeline("zero-shot-classification", model = "facebook/bart-large-mnli")
labels = ["good","bad"]

import pandas as pd
data = pd.read_csv("/", header = None,error_bad_lines=False)
data.columns = ["actual_labels","document"]