Naji20/sst2-sentiment-model

DistilBERT SST-2 Sentiment Analysis

This model is a fine-tuned version of distilbert-base-uncased on the SST-2 (Stanford Sentiment Treebank) dataset for sentiment analysis.

Model Description

• Model: DistilBERT (base, uncased)
• Task: Binary sentiment classification (Positive/Negative)
• Language: English
• Training Dataset: SST-2 (GLUE benchmark)
• Author: Naji (Аммар Нажи Али Саад)
• Institution: Kazan Federal University

Training Details

Training Data

• Dataset: GLUE SST-2
• Training samples: 67,349
• Validation samples: 872
• Test samples: 1,821

Training Configuration

• Epochs: 3
• Batch size: 16
• Learning rate: 2e-5
• Optimizer: AdamW
• Weight decay: 0.01
• Platform: Google Colab (T4 GPU)

Results

• Validation Accuracy: 89.9%
• Training Time: ~36 minutes on T4 GPU

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

# Load model and tokenizer
model_name = "Naji20/sst2-sentiment-model"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)

# Predict sentiment
text = "This movie is amazing!"
inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)
outputs = model(**inputs)
prediction = torch.argmax(outputs.logits, dim=1).item()
sentiment = "Positive" if prediction == 1 else "Negative"
print(f"Sentiment: {sentiment}")

Example Predictions

Text	Prediction
"This movie is absolutely amazing!"	Positive ✅
"I hated every minute of it."	Negative ❌
"Best film I've ever seen!"	Positive ✅
"Terrible waste of time."	Negative ❌

Limitations

• This model is trained only on English movie reviews
• Performance may vary on other domains (e.g., product reviews, tweets)
• Binary classification only (Positive/Negative, no neutral class)

Citation

@inproceedings{socher2013recursive,
  title={Recursive deep models for semantic compositionality over a sentiment treebank},
  author={Socher, Richard and Perelygin, Alex and Wu, Jean and Chuang, Jason and Manning, Christopher D and Ng, Andrew and Potts, Christopher},
  booktitle={Proceedings of the 2013 conference on empirical methods in natural language processing},
  pages={1631--1642},
  year={2013}
}

Model Card Authors

Naji (Аммар Нажи Али Саад) - Kazan Federal University