learning_rate: 0.0003
train_batch_size: 2
eval_batch_size: 6
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 4
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: 100
num_epochs: 5
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
0.3856	0.8282	200	1.6541	0.2701	0.6535
0.3525	1.6542	400	1.4525	0.2581	0.6213
0.2383	2.4803	600	1.5910	0.2510	0.6115
0.196	3.3064	800	1.7789	0.2420	0.5981
0.1448	4.1325	1000	1.7666	0.2403	0.5898
0.11	4.9607	1200	1.8025	0.2389	0.5832

ctaguchi/ssc-aln-xlsr300m-model-mix-adapt-max-lowlr

ssc-aln-xlsr300m-model-mix-adapt-max-lowlr