<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
mbart_cycle0_ko-en
This model is a fine-tuned version of reduced mbart-large-cc25(https://huggingface.co/facebook/mbart-large-cc25) on an custom dataset. It achieves the following results on the evaluation set:
- Loss: 8.0362
- Bleu: 3.9193
- Gen Len: 19.5758
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: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- total_train_batch_size: 16
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 300
- num_epochs: 50
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
|---|---|---|---|---|---|
| 8.5105 | 10.0 | 500 | 5.7366 | 1.0483 | 32.2222 |
| 1.3079 | 20.0 | 1000 | 7.1497 | 3.8281 | 17.3838 |
| 0.179 | 30.0 | 1500 | 7.7171 | 4.1437 | 18.6869 |
| 0.0535 | 40.0 | 2000 | 7.9881 | 4.1251 | 18.5455 |
| 0.0203 | 50.0 | 2500 | 8.0362 | 3.9193 | 19.5758 |
Framework versions
- Transformers 4.33.1
- Pytorch 2.0.1+cu117
- Datasets 2.14.5
- Tokenizers 0.13.3
