<!-- 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. -->
my_awesome_opus_books_model
This model is a fine-tuned version of t5-small on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.7068
- Bleu: 18.7799
- Gen Len: 15.8006
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: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
---|---|---|---|---|---|
2.3877 | 1.0 | 1250 | 2.1378 | 12.0972 | 15.964 |
2.2843 | 2.0 | 2500 | 2.0580 | 13.53 | 15.9109 |
2.2341 | 3.0 | 3750 | 1.9979 | 14.3797 | 15.9101 |
2.1456 | 4.0 | 5000 | 1.9475 | 14.9564 | 15.8565 |
2.1174 | 5.0 | 6250 | 1.9083 | 15.6059 | 15.8881 |
2.0813 | 6.0 | 7500 | 1.8743 | 16.1015 | 15.906 |
2.0502 | 7.0 | 8750 | 1.8449 | 16.6439 | 15.8066 |
2.0057 | 8.0 | 10000 | 1.8208 | 16.9371 | 15.8399 |
1.9874 | 9.0 | 11250 | 1.7971 | 17.3101 | 15.8339 |
1.9591 | 10.0 | 12500 | 1.7803 | 17.5769 | 15.8081 |
1.949 | 11.0 | 13750 | 1.7656 | 17.8263 | 15.7695 |
1.934 | 12.0 | 15000 | 1.7531 | 18.1862 | 15.7996 |
1.9163 | 13.0 | 16250 | 1.7419 | 18.3336 | 15.7962 |
1.9026 | 14.0 | 17500 | 1.7317 | 18.4599 | 15.7972 |
1.8914 | 15.0 | 18750 | 1.7255 | 18.5804 | 15.7754 |
1.8884 | 16.0 | 20000 | 1.7169 | 18.6063 | 15.8268 |
1.8975 | 17.0 | 21250 | 1.7134 | 18.6702 | 15.7824 |
1.8815 | 18.0 | 22500 | 1.7092 | 18.762 | 15.8065 |
1.8727 | 19.0 | 23750 | 1.7080 | 18.7832 | 15.7955 |
1.8733 | 20.0 | 25000 | 1.7068 | 18.7799 | 15.8006 |
Framework versions
- Transformers 4.29.2
- Pytorch 2.0.1+cu118
- Datasets 2.12.0
- Tokenizers 0.13.3