You can deploy this model and start earning money today!

devforam/
bert-sentiment-detector

Name: bert-sentiment-detector
Rating: 2.5 (5 reviews)

safetensors

bert

Views

No views yet

BERT Sentiment Detector

A sentiment analysis model built using BERT (bert-base-uncased) and fine-tuned on a labeled text dataset. The model classifies input text into sentiment categories using the Hugging Face Transformers framework.

Model Details

Model Architecture: BERT (bert-base-uncased)
Framework: Hugging Face Transformers
Task: Text Classification / Sentiment Detection
Training Library: Transformers Trainer API
Tokenization: AutoTokenizer from Hugging Face

Dataset

The model was trained on a dataset containing:

Text column: text
Label column: label
Label name column: label_name

Data preprocessing steps included:

Handling missing values
Label distribution analysis
Stratified train-test split
Tokenization using BERT tokenizer

Dataset split:

Train: 70%
Validation: 10%
Test: 20%

Training Configuration

Parameter	Value
Epochs	2
Learning Rate	2e-5
Batch Size	64
Optimizer	AdamW
Evaluation Strategy	Per Epoch
Weight Decay	0.01

Training was done using the Hugging Face Trainer API.

Model Performance

Evaluation metrics were calculated on the test set using:

Accuracy
Classification Report
Confusion Matrix

Example evaluation code:

python

from sklearn.metrics import classification_report
print(classification_report(y_true, y_pred))

How to Use the Model

Install dependencies

pip install transformers torch

Load Model

python

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="ForamSoni/bert-sentiment-detector"
)

classifier("I am feeling great today!")

Example Predictions

python

texts = [
    "I am super happy today!",
    "I feel terrible",
    "This is amazing",
    "I am feeling low"
]

classifier(texts)

Custom Prediction Function

python

def get_prediction(text):
    input_encoded = tokenizer(text, return_tensors='pt').to(device)

    with torch.no_grad():
        outputs = model(**input_encoded)

    logits = outputs.logits
    pred = torch.argmax(logits, dim=1).item()

    return id2label[pred]

Training Notebook

The model was trained using a Python notebook that includes:

Data analysis
Tokenization
Model training
Evaluation
Confusion matrix visualization
Inference pipeline

Project Structure

bert-sentiment-detector
│
├── model.safetensors
├── config.json
├── tokenizer.json
├── tokenizer_config.json
├── README.md
└── training.ipynb

Future Improvements

Train for more epochs
Larger dataset
Hyperparameter tuning
Deploy via API

Author

Foram Soni

Machine Learning Project using Hugging Face Transformers and BERT.

devforam/bert-sentiment-detector

BERT Sentiment Detector

Model Details

Dataset

Training Configuration

Model Performance

How to Use the Model

Install dependencies

Load Model

Example Predictions

Custom Prediction Function

Training Notebook

Project Structure

Future Improvements

Author

devforam/
bert-sentiment-detector