<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
bert-base-parsbert-uncased-finetuned-conditioned-khorshid
This model is a fine-tuned version of HooshvareLab/bert-base-parsbert-uncased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 4.1747
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 4.5994 | 1.0 | 324 | 4.3930 |
| 4.1349 | 2.0 | 648 | 4.2090 |
| 3.9812 | 3.0 | 972 | 4.0282 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2
