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
0.2347	0.4520	200	0.8176	0.2340	0.8195
0.2593	0.9040	400	0.7597	0.2228	0.7371
0.2031	1.3548	600	0.7099	0.2162	0.7250
0.194	1.8068	800	0.7108	0.2214	0.7219
0.1345	2.2576	1000	0.7149	0.2102	0.7112
0.14	2.7096	1200	0.6829	0.1985	0.7071
0.1112	3.1605	1400	0.7094	0.1937	0.7047
0.0854	3.6124	1600	0.7109	0.1819	0.6957
0.0721	4.0633	1800	0.7103	0.1803	0.6825
0.0658	4.5153	2000	0.7324	0.1780	0.6812
0.0552	4.9672	2200	0.7275	0.1794	0.6768

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

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