metadata
language:
- en
- he
tags:
- generated_from_trainer
model-index:
- name: marian_base
results: []
marian_base
This model is a fine-tuned version of orendar/en_he_base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.6365
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: 0.0001
- train_batch_size: 160
- eval_batch_size: 160
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 320
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 20.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.125 | 1.0 | 30633 | 2.0720 |
| 1.9288 | 2.0 | 61266 | 1.9033 |
| 1.8387 | 3.0 | 91899 | 1.8330 |
| 1.7832 | 4.0 | 122532 | 1.7864 |
| 1.7445 | 5.0 | 153165 | 1.7592 |
| 1.7152 | 6.0 | 183798 | 1.7404 |
| 1.6933 | 7.0 | 214431 | 1.7208 |
| 1.6743 | 8.0 | 245064 | 1.7005 |
| 1.6561 | 9.0 | 275697 | 1.6907 |
| 1.6431 | 10.0 | 306330 | 1.6903 |
| 1.6282 | 11.0 | 336963 | 1.6801 |
| 1.6173 | 12.0 | 367596 | 1.6714 |
| 1.6061 | 13.0 | 398229 | 1.6634 |
| 1.5971 | 14.0 | 428862 | 1.6543 |
| 1.5867 | 15.0 | 459495 | 1.6488 |
| 1.5781 | 16.0 | 490128 | 1.6447 |
| 1.5684 | 17.0 | 520761 | 1.6388 |
| 1.5597 | 18.0 | 551394 | 1.6416 |
| 1.5521 | 19.0 | 582027 | 1.6370 |
| 1.5438 | 20.0 | 612660 | 1.6365 |
Framework versions
- Transformers 4.18.0.dev0
- Pytorch 1.11.0+cu102
- Datasets 1.18.4
- Tokenizers 0.11.6