yesj1234/mbart_cycle1_ko-en

tst-translation-output

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: 3.7663
• Bleu: 19.3382
• Gen Len: 17.8929

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: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• total_eval_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
2.6161	11.09	2000	3.1762	13.5109	19.1966
2.6161	13.86	2500	3.0375	16.2868	18.7985
1.4467	16.62	3000	3.1328	17.6991	18.1949
1.4467	19.39	3500	3.2690	17.9052	18.3117
0.6809	22.15	4000	3.3850	18.4075	18.2149
0.6809	24.91	4500	3.4465	19.0339	18.009
0.3422	27.68	5000	3.5680	18.7281	17.5902
0.3422	30.44	5500	3.6350	19.1534	18.2177
0.1941	33.2	6000	3.7153	19.2575	17.8784
0.1941	35.97	6500	3.7382	19.2475	17.9831
0.1271	38.73	7000	3.7573	19.3045	17.9889

Framework versions

• Transformers 4.33.1
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3