t5-base-finetuned-pubmed
This model is a fine-tuned version of t5-base on the pub_med_summarization_dataset dataset. It achieves the following results on the evaluation set:
- Loss: 2.6311
- Rouge1: 9.3771
- Rouge2: 3.7042
- Rougel: 8.4912
- Rougelsum: 9.0013
- 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: 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: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|---|---|---|---|---|---|---|---|---|
| 2.0957 | 1.0 | 4000 | 1.9006 | 8.6968 | 3.2473 | 7.9565 | 8.3224 | 19.0 |
| 2.0489 | 2.0 | 8000 | 1.8571 | 8.6877 | 3.2461 | 7.9311 | 8.2991 | 19.0 |
| 2.7345 | 3.0 | 12000 | 2.6112 | 9.585 | 3.0129 | 8.4729 | 9.1109 | 19.0 |
| 3.0585 | 4.0 | 16000 | 2.7222 | 9.7011 | 3.3549 | 8.6588 | 9.2646 | 19.0 |
| 2.9437 | 5.0 | 20000 | 2.6311 | 9.3771 | 3.7042 | 8.4912 | 9.0013 | 19.0 |
Framework versions
- Transformers 4.16.2
- Pytorch 1.10.0+cu111
- Datasets 1.18.3
- Tokenizers 0.11.6