umT5-joker-finetuned-en-es-fr
This model is a fine-tuned version of google/umt5-base on the CLEF JOKER 2023 dataset. It achieves the following results on the evaluation set:
- Loss: 0.6818
- Accuracy: 0.5679
- F1: 0.6191
- Precision: 0.5495
- Recall: 0.7088
<!--
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: 16
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall |
|---|---|---|---|---|---|---|---|
| 0.7081 | 1.0 | 611 | 0.7098 | 0.5007 | 0.6518 | 0.4979 | 0.9433 |
| 0.7079 | 2.0 | 1222 | 0.6955 | 0.5255 | 0.6312 | 0.5132 | 0.8195 |
| 0.6973 | 3.0 | 1833 | 0.6818 | 0.5679 | 0.6191 | 0.5495 | 0.7088 |
| 0.6869 | 4.0 | 2444 | 0.6904 | 0.5561 | 0.6256 | 0.5374 | 0.7484 |
| 0.6837 | 5.0 | 3055 | 0.6835 | 0.5529 | 0.6291 | 0.5340 | 0.7655 |
Framework versions
- Transformers 4.33.2
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.13.3
