metadata
tags:
- generated_from_trainer
metrics:
- bleu
model-index:
- name: M2M100-ja-zh
results: []
M2M100-ja-zh
This model is a fine-tuned version of facebook/m2m100_418M on the larryvrh/CCMatrix-v1-Ja_Zh-filtered and Nekofox/ja-zh-twitter-translate dataset. It achieves the following results on the evaluation set:
- Loss: 1.6727
- Bleu: 29.1635
- Gen Len: 19.2263
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: 4e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine_with_restarts
- lr_scheduler_warmup_ratio: 0.07
- num_epochs: 1
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
|---|---|---|---|---|---|
| 1.743 | 1.0 | 14178 | 1.6727 | 29.1635 | 19.2263 |
Framework versions
- Transformers 4.28.1
- Pytorch 2.0.0+cu118
- Datasets 2.12.0
- Tokenizers 0.13.3