BertForSequenceClassification on CNN News Dataset

This repository contains a fine-tuned Bert base model for sequence classification on the CNN News dataset. The model is able to classify news articles into one of six categories: business, entertainment, health, news, politics, and sport.

The model was fine-tuned for four epochs achieving a training loss of 0.077900, a validation loss of 0.190814

• accuracy : 0.956690.
• f1 : 0.956144.
• precision : 0.956393
• recall : 0.956690

Model Description

• Developed by: CHERGUELAINE Ayoub
• Shared by : HuggingFace
• Model type: Language model
• Language(s) (NLP): en
• Finetuned from model : distilbert-base-uncased

Usage

You can use this model with the Hugging Face Transformers library for a variety of natural language processing tasks, such as text classification, sentiment analysis, and more.

Here's an example of how to use this model for text classification in Python:

from transformers import AutoTokenizer, BertForSequenceClassification

model_name = "AyoubChLin/bert_cnn_news"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = TFAutoModelForSequenceClassification.from_pretrained(model_name)

text = "This is a news article about politics."
inputs = tokenizer(text, padding=True, truncation=True, return_tensors="tf")

outputs = model(inputs)
predicted_class_id = tf.argmax(outputs.logits, axis=-1).numpy()[0]

labels = ["business", "entertainment", "health", "news", "politics", "sport"]
predicted_label = labels[predicted_class_id]

In this example, we first load the tokenizer and the model using their respective from_pretrained methods. We then encode a news article using the tokenizer, pass the inputs through the model, and extract the predicted label using the argmax function. Finally, we map the predicted label to its corresponding category using a list of labels.

Contributors

This model was fine-tuned by CHERGUELAINE Ayoub.