| license: apache-2.0 | |
| library_name: transformers | |
| tags: | |
| - sweep | |
| - fine-tuned | |
| # SweepBestModel | |
| This model was selected as the best checkpoint from a hyperparameter sweep over 8 runs. It achieved the highest evaluation accuracy across all configurations tested. | |
| ## Model Details | |
| - **Model type:** Encoder-only transformer | |
| - **Training objective:** Sequence classification | |
| - **Sweep strategy:** Bayesian optimization | |
| ## Evaluation Results | |
| | Metric | Value | | |
| |---|---| | |
| | eval_accuracy | 0.901 | | |
| | eval_loss | 0.245 | | |
| | f1_score | 0.894 | | |
| | precision | 0.908 | | |
| | recall | 0.881 | | |
| ## Training Configuration | |
| The best run was trained with the following hyperparameters: | |
| | Parameter | Value | | |
| |---|---| | |
| | learning_rate | 2e-05 | | |
| | batch_size | 64 | | |
| | epochs | 12 | | |
| | weight_decay | 0.001 | | |
| ## Usage | |
| ```python | |
| from transformers import AutoModelForSequenceClassification, AutoTokenizer | |
| model = AutoModelForSequenceClassification.from_pretrained("SweepBestModel-TestRepo") | |
| tokenizer = AutoTokenizer.from_pretrained("SweepBestModel-TestRepo") | |
| ``` | |
| ## License | |
| This model is licensed under the Apache-2.0 License. | |