AspectLens DistilBERT ABSA Classifier
This model is a fine-tuned distilbert-base-uncased sequence classifier for aspect-conditioned sentiment classification on restaurant review aspects.
Input format:
review: The food was excellent, but the service was slow. [SEP] aspect: service
Labels:
- negative
- neutral
- positive
- conflict
Held-out evaluation results from the project split:
| Metric | Score |
|---|---|
| Accuracy | 0.783 |
| Macro-F1 | 0.541 |
| Weighted-F1 | 0.771 |
The model was created for the PROG74040 Phase II course project prototype. The rare conflict class remains the weakest class due to limited support.
- Downloads last month
- 49
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support