metadata
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- bleu
model-index:
- name: MarianMix_en-10
results: []
MarianMix_en-10
This model is a fine-tuned version of Helsinki-NLP/opus-tatoeba-en-ja on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.6475
- Bleu: 3.3068
- Gen Len: 46.466
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: 32
- eval_batch_size: 32
- seed: 99
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 10
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
|---|---|---|---|---|---|
| 1.7048 | 0.44 | 500 | 1.0358 | 0.7054 | 50.216 |
| 0.6727 | 0.89 | 1000 | 0.8751 | 1.1453 | 46.401 |
| 0.5497 | 1.33 | 1500 | 0.7978 | 1.2939 | 46.27 |
| 0.5087 | 1.77 | 2000 | 0.7424 | 1.7606 | 48.227 |
| 0.4488 | 2.22 | 2500 | 0.7176 | 2.1927 | 47.076 |
| 0.4077 | 2.66 | 3000 | 0.6888 | 2.5931 | 46.162 |
| 0.3895 | 3.1 | 3500 | 0.6776 | 2.975 | 45.728 |
| 0.3465 | 3.55 | 4000 | 0.6645 | 2.9679 | 46.047 |
| 0.3464 | 3.99 | 4500 | 0.6533 | 3.435 | 46.317 |
| 0.3119 | 4.43 | 5000 | 0.6517 | 3.4986 | 46.333 |
| 0.3066 | 4.88 | 5500 | 0.6475 | 3.3068 | 46.466 |
Framework versions
- Transformers 4.12.5
- Pytorch 1.9.1
- Datasets 1.17.0
- Tokenizers 0.10.3