distilbert-imdb-sentiment-graduation
This is a fine-tuned version of distilbert-base-uncased on the IMDB sentiment dataset.
Model description
This model classifies movie reviews as positive or negative.
Training and evaluation data
- Dataset: IMDB (25,000 training samples)
- Epochs: 3 (Reverted to Epoch 2 checkpoint to prevent overfitting)
- Learning Rate: 2e-5
Evaluation Results
Based on the validation metrics, we selected the checkpoint at step 6000 (Epoch 2).
- Accuracy: 0.93+
- Precision: ~0.93
- Recall: ~0.94
- F1 Score: ~0.93
How to use
from transformers import pipeline
classifier = pipeline("sentiment-analysis", model="YOUR_USERNAME/distilbert-imdb-sentiment-graduation")
classifier("This movie was fantastic!")
- Downloads last month
- 6