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: 1.1639
- Bleu: 15.2673
- Gen Len: 50.7578
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 |
|---|---|---|---|---|---|
| 2.2528 | 0.44 | 500 | 2.3059 | 0.2046 | 68.5206 |
| 1.2562 | 0.89 | 1000 | 1.9629 | 1.2393 | 61.1236 |
| 1.0764 | 1.33 | 1500 | 1.6859 | 4.1958 | 57.1075 |
| 0.95 | 1.78 | 2000 | 1.5075 | 6.683 | 55.8533 |
| 0.8486 | 2.22 | 2500 | 1.3848 | 9.3923 | 53.5568 |
| 0.7887 | 2.67 | 3000 | 1.3047 | 11.7962 | 53.6583 |
| 0.7488 | 3.11 | 3500 | 1.2574 | 12.8451 | 54.3196 |
| 0.692 | 3.56 | 4000 | 1.2154 | 13.6083 | 51.0131 |
| 0.6795 | 4.0 | 4500 | 1.1881 | 14.7516 | 51.6563 |
| 0.6375 | 4.45 | 5000 | 1.1714 | 15.1401 | 51.5437 |
| 0.6332 | 4.89 | 5500 | 1.1639 | 15.2673 | 50.7578 |
Framework versions
- Transformers 4.12.5
- Pytorch 1.9.1
- Datasets 1.17.0
- Tokenizers 0.10.3