learning_rate: 3e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 64
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: 400
num_epochs: 24
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Wer
0.4779	1.0	119	0.1638	0.0073	0.0939
0.1553	2.0	238	0.1157	0.0073	0.0864
0.1239	3.0	357	0.1218	0.0073	0.1068
0.1109	4.0	476	0.1010	0.0073	0.0682
0.087	5.0	595	0.0852	0.0073	0.1305
0.0715	6.0	714	0.0624	0.0073	0.0581
0.0537	7.0	833	0.0609	0.0073	0.0842
0.044	8.0	952	0.0456	0.0073	0.0374
0.0348	9.0	1071	0.0357	0.0073	0.0245
0.0286	10.0	1190	0.0270	0.0073	0.0148
0.024	11.0	1309	0.0127	0.0073	0.0090
0.0179	12.0	1428	0.0240	0.0073	0.0090
0.0136	13.0	1547	0.0076	0.0073	0.0073
0.0097	14.0	1666	0.0048	0.0073	0.0032
0.0091	15.0	1785	0.0057	0.0073	0.0016
0.0065	16.0	1904	0.0058	0.0073	0.0048
0.0036	17.0	2023	0.0013	0.0073	0.0022
0.0023	18.0	2142	0.0002	0.0073	0.0
0.001	19.0	2261	0.0002	0.0073	0.0
0.0006	20.0	2380	0.0001	0.0073	0.0
0.0005	21.0	2499	0.0001	0.0073	0.0
0.0002	22.0	2618	0.0001	0.0073	0.0
0.0001	23.0	2737	0.0001	0.0073	0.0
0.0001	23.8042	2832	0.0001	0.0073	0.0

dhaval0108/ap-XEZzYyyLLduYPqqPZLCHwP

ap-XEZzYyyLLduYPqqPZLCHwP