t5-small-normal-arxiv
This model is a fine-tuned version of t5-small on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 2.6373
- Rouge1: 0.1728
- Rouge2: 0.0617
- Rougel: 0.1378
- Rougelsum: 0.1377
- Gen Len: 19.0
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: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|---|---|---|---|---|---|---|---|---|
| No log | 1.0 | 403 | 2.7717 | 0.1646 | 0.0557 | 0.1319 | 0.1319 | 19.0 |
| 3.0546 | 2.0 | 806 | 2.7195 | 0.1684 | 0.0585 | 0.1347 | 0.1347 | 19.0 |
| 2.8771 | 3.0 | 1209 | 2.6899 | 0.1695 | 0.0596 | 0.1356 | 0.1356 | 19.0 |
| 2.8364 | 4.0 | 1612 | 2.6719 | 0.1716 | 0.0605 | 0.137 | 0.1369 | 19.0 |
| 2.8058 | 5.0 | 2015 | 2.6585 | 0.1717 | 0.061 | 0.1371 | 0.1371 | 19.0 |
| 2.8058 | 6.0 | 2418 | 2.6504 | 0.1721 | 0.0615 | 0.1374 | 0.1374 | 19.0 |
| 2.7852 | 7.0 | 2821 | 2.6453 | 0.1725 | 0.0618 | 0.1378 | 0.1377 | 19.0 |
| 2.778 | 8.0 | 3224 | 2.6404 | 0.1728 | 0.0618 | 0.1378 | 0.1378 | 19.0 |
| 2.7612 | 9.0 | 3627 | 2.6386 | 0.1724 | 0.0615 | 0.1375 | 0.1375 | 19.0 |
| 2.7644 | 10.0 | 4030 | 2.6373 | 0.1728 | 0.0617 | 0.1378 | 0.1377 | 19.0 |
Framework versions
- Transformers 4.45.1
- Pytorch 2.4.0
- Datasets 3.0.1
- Tokenizers 0.20.0