learning_rate: 5e-05
train_batch_size: 64
eval_batch_size: 64
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 128
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 300
num_epochs: 100
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Bleu-4
No log	1.0	118	1.4441	0.1228
No log	2.0	236	1.3026	0.1406
2.0335	3.0	354	1.2238	0.1685
2.0335	4.0	472	1.1742	0.1841
2.0335	5.0	590	1.1522	0.1944
1.0795	6.0	708	1.1471	0.2103
1.0795	7.0	826	1.1434	0.2125
0.8075	8.0	944	1.1574	0.2108
0.8075	9.0	1062	1.1881	0.2121

Skyler215/SwinV2_Syllable

SwinV2_Syllable