flan-t5-small-compression
This model is a fine-tuned version of google/flan-t5-small on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5181
- Rouge1: 0.8820
- Rouge2: 0.7104
- Rougel: 0.8485
- Rougelsum: 0.8488
- Comp Ratio Mean: 0.6611
- Comp Ratio P90: 0.7674
- Pct Violations: 0.0
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: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Use adafactor and the args are: No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Comp Ratio Mean | Comp Ratio P90 | Pct Violations |
|---|---|---|---|---|---|---|---|---|---|---|
| 1.2576 | 1.0 | 1594 | 0.6457 | 0.8528 | 0.6587 | 0.8197 | 0.8199 | 0.6626 | 0.7736 | 0.0 |
| 0.7688 | 2.0 | 3188 | 0.5727 | 0.8689 | 0.6851 | 0.8345 | 0.8349 | 0.6647 | 0.7694 | 0.0 |
| 0.6591 | 3.0 | 4782 | 0.5405 | 0.8750 | 0.6963 | 0.8413 | 0.8417 | 0.6684 | 0.7692 | 0.0 |
| 0.5957 | 4.0 | 6376 | 0.5333 | 0.8771 | 0.7002 | 0.8438 | 0.8440 | 0.6600 | 0.7660 | 0.0 |
| 0.548 | 5.0 | 7970 | 0.5212 | 0.8792 | 0.7059 | 0.8467 | 0.8470 | 0.6617 | 0.7648 | 0.0004 |
| 0.5139 | 6.0 | 9564 | 0.5196 | 0.8799 | 0.7064 | 0.8472 | 0.8473 | 0.6597 | 0.7636 | 0.0 |
| 0.4862 | 7.0 | 11158 | 0.5144 | 0.8805 | 0.7076 | 0.8473 | 0.8474 | 0.6656 | 0.7705 | 0.0004 |
| 0.466 | 8.0 | 12752 | 0.5157 | 0.8819 | 0.7098 | 0.8489 | 0.8492 | 0.6622 | 0.7674 | 0.0 |
| 0.4499 | 9.0 | 14346 | 0.5156 | 0.8816 | 0.7096 | 0.8486 | 0.8489 | 0.6604 | 0.7660 | 0.0 |
| 0.4393 | 10.0 | 15940 | 0.5181 | 0.8820 | 0.7104 | 0.8485 | 0.8488 | 0.6611 | 0.7674 | 0.0 |
Framework versions
- Transformers 4.57.1
- Pytorch 2.6.0+cu124
- Datasets 4.4.1
- Tokenizers 0.22.1