Sentiment Classification

Fine-tuning a BERT model (Huggingface API) for Sentiment Classification on the SMILE: Twitter Emotion Dataset

System Information [Trained Using Colab]

• torch : 1.5.0
• pandas : 1.0.3
• numpy : 1.18.4

Results

• F1 Score (Weighted) : 0.853
• Accuracy Per Class (See Notebook)
• Training time : 90 seconds per Epoch (approximately)

References

• Huggingface API
• Jay Alammar's The Illustrated BERT
• A Gentle Visual Intro to Data Analysis in Python Using Pandas
• SMILE Twitter Emotion Dataset
• Classification Accuracy is Not Enough