learning_rate: 0.0001
train_batch_size: 2
eval_batch_size: 4
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
lr_scheduler_warmup_steps: 500
training_steps: 100000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.3379	3.9062	5000	0.7792	47.4376
0.1422	7.8125	10000	0.7314	38.2661
0.1088	11.7188	15000	0.7569	38.1037
0.0704	15.625	20000	0.7331	37.7788
0.0265	19.5312	25000	0.7670	39.1670
0.0179	23.4375	30000	0.7927	36.2428
0.0225	27.3438	35000	0.7842	35.8883
0.0227	31.25	40000	0.7847	34.4705
0.0167	35.1562	45000	0.7585	32.4472
0.0149	39.0625	50000	0.7667	31.9894
0.0288	42.9688	55000	0.7391	32.6983
0.0067	46.875	60000	0.7365	31.1475
0.0048	50.7812	65000	0.7247	29.8626
0.0184	54.6875	70000	0.7135	30.5272
0.0051	58.5938	75000	0.7035	29.5525
0.0185	62.5	80000	0.6938	29.1833
0.0008	66.4062	85000	0.6732	30.0399
0.0068	70.3125	90000	0.6482	28.3710
0.0065	74.2188	95000	0.6310	27.7212
0.0001	78.125	100000	0.6338	26.8203

Eimhin03/Shunya_Augmentation_enabled_attempt2

Shunya_Augmentation_enabled_attempt2