iTzMiNOS/BERT-finetuned-multiclass-tweet-sentiment-analysis

Overview

This repository contains a fine-tuned BERT (bert-base-uncased) model for multi-class sentiment analysis of tweets. The model was trained on a labeled dataset of tweets, allowing it to classify text into different sentiment categories.

Model Details

Base Model: bert-base-uncased

Dataset: Custom twitter labeled data

Task: Multi-class sentiment analysis

Fine-Tuning Framework: Hugging Face transformers

Evaluation Metrics: Accuracy, F1-score

Training Process

Data Preparation:

The dataset was split into training (70%), validation (10%), and test (20%) sets.

Text data was tokenized using AutoTokenizer from Hugging Face.

Model Configuration:

Fine-tuned BERT-base-uncased with classification head.

label2id and id2label mappings created for multi-class classification.

Mappings

['joy', 'anger', 'sadness', 'surprised', 'love', 'fear']

Training Parameters:

Epochs: 2

Batch Size: 64

Learning Rate: 2e-5

Weight Decay: 0.01

Evaluation:

Used accuracy and weighted F1-score as performance metrics.

Generated a classification report and confusion matrix.

Inference:

Text classification pipeline created using Hugging Face pipeline.

Example prediction:

text = "I am feeling happy today!"
classifier = pipeline('text-classification', model='iTzMiNOS/BERT-finetuned-multiclass-tweet-sentiment-analysis')
print(classifier(text))

Installation & Usage

Install Dependencies

pip install transformers datasets torch

Load the Model

from transformers import pipeline

classifier = pipeline('text-classification', model='iTzMiNOS/BERT-finetuned-multiclass-tweet-sentiment-analysis')

text = "I am so excited for the weekend!"
result = classifier(text)
print(result)

Evaluate Model Performance

from sklearn.metrics import classification_report, confusion_matrix
import seaborn as sns
import matplotlib.pyplot as plt
import numpy as np

# Example test set predictions

y_true = [...]  # Actual labels
y_pred = [...]  # Model predictions

print(classification_report(y_true, y_pred))

cm = confusion_matrix(y_true, y_pred)
plt.figure(figsize=(5,5))
sns.heatmap(cm, annot=True, fmt='d', cmap='Reds')
plt.ylabel("Actual")
plt.xlabel("Predicted")
plt.show()

License

This model is open-source and can be used under the MIT License.