learning_rate: 2e-05
train_batch_size: 64
eval_batch_size: 64
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 15
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Bleu
2.1935	1.0	1167	1.8939	12.2773
1.8284	2.0	2334	1.6966	19.6177
1.6192	3.0	3501	1.6043	23.4603
1.5213	4.0	4668	1.5553	24.4698
1.44	5.0	5835	1.5231	25.0087
1.3675	6.0	7002	1.5010	26.2754
1.3185	7.0	8169	1.4864	26.7454
1.2766	8.0	9336	1.4758	27.1840
1.2286	9.0	10503	1.4652	27.3690
1.2045	10.0	11670	1.4613	27.5355
1.1712	11.0	12837	1.4573	28.0098
1.1546	12.0	14004	1.4546	27.9113
1.1281	13.0	15171	1.4537	28.2765
1.1166	14.0	16338	1.4528	28.3006
1.1113	15.0	17505	1.4522	28.3689

NetherQuartz/tatoeba-tok-vi

tatoeba-tok-vi