<!-- 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. -->
flan-t5-base-SQuAD-QG
This model is a fine-tuned version of google/flan-t5-base on the qg_squad dataset. It achieves the following results on the evaluation set:
- Loss: 0.5526
- Rouge1: 52.5686
- Rouge2: 29.8958
- Rougel: 48.6057
- Rougelsum: 48.6025
- Meteor: 47.5477
- Bleu-n: 21.1573
- Bleu-1: 53.1650
- Bleu-2: 27.0832
- Bleu-3: 17.3550
- Bleu-4: 11.6241
- Gen Len: 14.2079
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: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Meteor | Bleu-n | Bleu-1 | Bleu-2 | Bleu-3 | Bleu-4 | Gen Len |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.6194 | 1.0 | 2366 | 0.5616 | 51.4072 | 28.7617 | 47.4858 | 47.496 | 46.1739 | 20.4191 | 52.6343 | 26.3347 | 16.8839 | 11.3072 | 14.0406 |
| 0.5727 | 2.0 | 4733 | 0.5540 | 52.0791 | 29.3729 | 48.1664 | 48.1802 | 46.8836 | 20.7120 | 53.2071 | 26.9025 | 17.1791 | 11.4335 | 14.0766 |
| 0.55 | 3.0 | 7099 | 0.5521 | 52.2372 | 29.5935 | 48.2874 | 48.2968 | 47.1455 | 20.9160 | 52.9894 | 26.8627 | 17.2081 | 11.5033 | 14.1710 |
| 0.5293 | 4.0 | 9466 | 0.5523 | 52.5425 | 29.8483 | 48.558 | 48.5627 | 47.4331 | 21.0784 | 53.3308 | 27.1435 | 17.3826 | 11.6385 | 14.1482 |
| 0.5225 | 5.0 | 11830 | 0.5526 | 52.5686 | 29.8958 | 48.6057 | 48.6025 | 47.5477 | 21.1573 | 53.1650 | 27.0832 | 17.3550 | 11.6241 | 14.2079 |
Framework versions
- Transformers 4.28.1
- Pytorch 2.0.0
- Datasets 2.1.0
- Tokenizers 0.13.3
