learning_rate: 0.0001
train_batch_size: 20
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 40
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 500
training_steps: 3000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss
0.5944	0.2829	100	0.5223
0.4894	0.5658	200	0.4270
0.478	0.8487	300	0.4273
0.4693	1.1301	400	0.4180
0.4616	1.4130	500	0.4087
0.4391	1.6959	600	0.4041
0.4407	1.9788	700	0.3936
0.4328	2.2603	800	0.3948
0.4242	2.5431	900	0.3867
0.4252	2.8260	1000	0.3936
0.4098	3.1075	1100	0.3795
0.4128	3.3904	1200	0.3834
0.4098	3.6733	1300	0.3762
0.4034	3.9562	1400	0.3724
0.4003	4.2376	1500	0.3717
0.4018	4.5205	1600	0.3723
0.4027	4.8034	1700	0.3705
0.4005	5.0849	1800	0.3696
0.3923	5.3678	1900	0.3670
0.3885	5.6506	2000	0.3650
0.4018	5.9335	2100	0.3642
0.3836	6.2150	2200	0.3634
0.3865	6.4979	2300	0.3637
0.3821	6.7808	2400	0.3617
0.3881	7.0622	2500	0.3619
0.3834	7.3451	2600	0.3587
0.3858	7.6280	2700	0.3584
0.378	7.9109	2800	0.3571
0.3785	8.1924	2900	0.3568
0.381	8.4752	3000	0.3570

seywan1378/speecht5_tts_hataw_ckb

SpeechT5 TTS Hataw