Model description
The rotten-tomatoes-model is a text-classification model. It used the bert-base-cased model, and was fine tuned on the rotten_tomatoes model.
After inputting a movie review, the model will output its prediction of how positive/negative the review is. LABEL_0 is Negative, while LABEL_1 is Positive.
Intended uses & limitations
This model can be used to take in movie reviews and predict whether the overall sentiments of the review are positive or negative.
An example use case for this model is taking in reviews spanning from the start of the pandemic to the current time to see how sentiments surrounding movies might have been affected by when in the pandemic it was released (or other factors such as the method it was released).
Training and evaluation data
As mentioned above, this model was fine-tuned on the rotten_tomatoes dataset, which contains 5,331 positive and 5,331 negative movie reviews from Rotten Tomatoes.
Training results
| Train Loss | Train Accuracy | Validation Loss | Validation Accuracy | Epoch |
|---|
| 0.4028 | 0.8213 | 0.4626 | 0.8433 | 0 |
| 0.1628 | 0.9390 | 0.3498 | 0.8696 | 1 |
| 0.0386 | 0.9878 | 0.4790 | 0.8621 | 2 |
Framework versions
- Transformers 4.18.0
- TensorFlow 2.8.0
- Datasets 2.1.0
- Tokenizers 0.12.1
