mt5-base-finetuned-v1
This model is a fine-tuned version of google/mt5-base on the Persian News dataset. It achieves the following results on the evaluation set:
- Loss: 1.087988
- Rouge1: 1.2887
- Rouge2: 0.1861
- Rougel: 1.2862
- Rougelsum: 1.2818
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5.6e-05
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum |
|---|---|---|---|---|---|---|---|
| 1.223400 | 1 | 20437 | 1.153162 | 1.0624 | 0.1351 | 1.0668 | 1.0740 |
| 1.202900 | 2 | 40874 | 1.086163 | 1.1579 | 0.1426 | 1.1724 | 1.1599 |
| 1.173500 | 3 | 61311 | 1.087988 | 1.2887 | 0.1861 | 1.2862 | 1.2818 |
Framework versions
- Transformers 4.21.1
- Pytorch 1.12.0+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1