yesj1234/mbart_cycle1_ko-zh

<!-- 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. -->

tst-translation-output2

This model is a fine-tuned version of mbart-large-cc25 on an custom dataset. It achieves the following results on the evaluation set:

• —Loss: 3.4005
• —Bleu: 26.0229
• —Gen Len: 15.1659

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
• —trainbatchsize: 4
• —evalbatchsize: 4
• —seed: 42
• —distributed_type: multi-GPU
• —num_devices: 4
• —totaltrainbatch_size: 16
• —totalevalbatch_size: 16
• —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• —lrschedulertype: linear
• —lrschedulerwarmup_steps: 500
• —num_epochs: 50

Training results

... | 0.0042 | 40.42 | 35000 | 3.3485 | 25.2464 | 15.2387 | | 0.0029 | 41.57 | 36000 | 3.3744 | 25.2885 | 15.1306 | | 0.0026 | 42.73 | 37000 | 3.3947 | 25.9359 | 15.1896 | | 0.0024 | 43.88 | 38000 | 3.3699 | 25.5309 | 15.2671 | | 0.0022 | 45.03 | 39000 | 3.3947 | 25.2932 | 15.1387 | | 0.0011 | 46.19 | 40000 | 3.4075 | 25.7551 | 15.1231 | | 0.001 | 47.34 | 41000 | 3.3918 | 25.6345 | 15.1243 | | 0.0007 | 48.5 | 42000 | 3.4063 | 25.7209 | 15.111 | | 0.0006 | 49.65 | 43000 | 3.4003 | 25.9227 | 15.1873 |

Framework versions

• —Transformers 4.32.1
• —Pytorch 2.0.1+cu117
• —Datasets 2.14.4
• —Tokenizers 0.13.3