paraphraser-german-mt5-small
This model is a fine-tuned version of google/mt5-small on the paws-x (de) and tapaco (de) dataset. It achieves the following results on the evaluation set:
- Loss: 1.7678
- Perplexity: 5.86
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: 10
- eval_batch_size: 10
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 3.7064 | 0.05 | 2000 | 2.0731 |
| 2.8673 | 0.11 | 4000 | 2.0420 |
| 2.6133 | 0.16 | 6000 | 2.0080 |
| 2.4563 | 0.21 | 8000 | 1.9556 |
| 2.385 | 0.27 | 10000 | 1.9090 |
| 2.3122 | 0.32 | 12000 | 1.9127 |
| 2.2775 | 0.38 | 14000 | 1.8658 |
| 2.2323 | 0.43 | 16000 | 1.8407 |
| 2.17 | 0.48 | 18000 | 1.8342 |
| 2.1672 | 0.54 | 20000 | 1.8328 |
| 2.1488 | 0.59 | 22000 | 1.8071 |
| 2.1026 | 0.64 | 24000 | 1.8328 |
| 2.1036 | 0.7 | 26000 | 1.7979 |
| 2.0854 | 0.75 | 28000 | 1.7895 |
| 2.0594 | 0.81 | 30000 | 1.7944 |
| 2.0793 | 0.86 | 32000 | 1.7726 |
| 2.0661 | 0.91 | 34000 | 1.7762 |
| 2.0722 | 0.97 | 36000 | 1.7714 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.13.1+cu116
- Datasets 2.9.0
- Tokenizers 0.13.2