| license: apache-2.0 | |
| library_name: transformers | |
| tags: | |
| - text-classification | |
| - roberta | |
| - sweep | |
| # SweepBestModel | |
| <div align="center"> | |
| <img src="figures/training_curve.png" width="70%" alt="Training Curve" /> | |
| </div> | |
| ## Overview | |
| This model was selected from a hyperparameter sweep as the best-performing run based on validation accuracy. It is a RoBERTa-based sequence classifier fine-tuned on our internal dataset. | |
| ## Training Configuration | |
| - **Learning Rate:** 3e-5 | |
| - **Batch Size:** 64 | |
| - **Epochs:** 10 | |
| - **Best Validation Accuracy:** 0.864 | |
| ## Benchmark Results | |
| <div align="center"> | |
| | Benchmark | Score | | |
| |---|---| | |
| | MNLI (m/mm) | 0.864 | | |
| | SST-2 | 0.864 | | |
| | QQP | 0.864 | | |
| | QNLI | 0.864 | | |
| | RTE | 0.864 | | |
| | CoLA | 0.864 | | |
| | STS-B | 0.864 | | |
| | MRPC | 0.864 | | |
| </div> | |
| ## Usage | |
| ```python | |
| from transformers import AutoModelForSequenceClassification, AutoTokenizer | |
| model = AutoModelForSequenceClassification.from_pretrained("SweepBestModel-Repo") | |
| tokenizer = AutoTokenizer.from_pretrained("SweepBestModel-Repo") | |
| ``` | |
| ## Figures | |
| <div align="center"> | |
| <img src="figures/confusion_matrix.png" width="45%" alt="Confusion Matrix" /> | |
| <img src="figures/loss_curve.png" width="45%" alt="Loss Curve" /> | |
| </div> |