mt5-bleu-durga-q1-clean
This model is a fine-tuned version of google/mt5-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.7819
- Bleu: 0.0359
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: 0.0003
- train_batch_size: 20
- eval_batch_size: 20
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu |
|---|---|---|---|---|
| 15.8442 | 1.0 | 3 | 11.1246 | 0.0 |
| 13.0661 | 2.0 | 6 | 9.3553 | 0.0 |
| 11.7048 | 3.0 | 9 | 8.0317 | 0.0 |
| 8.87 | 4.0 | 12 | 7.1382 | 0.0 |
| 11.0893 | 5.0 | 15 | 6.7905 | 0.0 |
| 9.8787 | 6.0 | 18 | 6.5255 | 0.0 |
| 9.8189 | 7.0 | 21 | 6.7007 | 0.0 |
| 8.2022 | 8.0 | 24 | 6.2109 | 0.0 |
| 8.5899 | 9.0 | 27 | 5.9520 | 0.0 |
| 7.5305 | 10.0 | 30 | 5.5748 | 0.0 |
| 7.0381 | 11.0 | 33 | 5.2219 | 0.0054 |
| 6.675 | 12.0 | 36 | 4.8006 | 0.0046 |
| 7.4134 | 13.0 | 39 | 4.3795 | 0.0051 |
| 5.8722 | 14.0 | 42 | 3.9322 | 0.0099 |
| 4.5875 | 15.0 | 45 | 3.5017 | 0.0079 |
| 5.3675 | 16.0 | 48 | 3.1927 | 0.0 |
| 4.2999 | 17.0 | 51 | 2.8956 | 0.0110 |
| 4.3349 | 18.0 | 54 | 2.7138 | 0.0088 |
| 3.9688 | 19.0 | 57 | 2.5350 | 0.0 |
| 4.2931 | 20.0 | 60 | 2.4138 | 0.0 |
| 3.8427 | 21.0 | 63 | 2.3127 | 0.0 |
| 3.2991 | 22.0 | 66 | 2.2054 | 0.0 |
| 3.1351 | 23.0 | 69 | 2.1069 | 0.0 |
| 3.023 | 24.0 | 72 | 2.0208 | 0.0 |
| 3.4366 | 25.0 | 75 | 1.9500 | 0.0272 |
| 2.7941 | 26.0 | 78 | 1.9068 | 0.0370 |
| 2.9454 | 27.0 | 81 | 1.8419 | 0.0365 |
| 2.6117 | 28.0 | 84 | 1.8775 | 0.0367 |
| 2.6785 | 29.0 | 87 | 1.7772 | 0.0361 |
| 2.7523 | 30.0 | 90 | 1.7819 | 0.0359 |
Framework versions
- Transformers 4.46.1
- Pytorch 2.5.0+cu121
- Datasets 3.0.2
- Tokenizers 0.20.1