Model description

More information needed

Intended uses & limitations

More information needed

More information needed

The following hyperparameters were used during training:

learning_rate: 1e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
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: constant_with_warmup
lr_scheduler_warmup_steps: 50
training_steps: 500
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
No log	3.3333	20	0.0829	14.2857	9.5238
0.3291	6.6667	40	0.0072	0.0	0.0
0.0256	10.0	60	0.0015	0.0	0.0
0.0009	13.3333	80	0.0010	0.0	0.0
0.0004	16.6667	100	0.0007	0.0	0.0
0.0004	20.0	120	0.0006	0.0	0.0
0.0003	23.3333	140	0.0005	0.0	0.0
0.0002	26.6667	160	0.0004	0.0	0.0
0.0002	30.0	180	0.0004	0.0	0.0
0.0001	33.3333	200	0.0004	0.0	0.0
0.0001	36.6667	220	0.0003	0.0	0.0
0.0001	40.0	240	0.0003	0.0	0.0
0.0001	43.3333	260	0.0003	0.0	0.0
0.0001	46.6667	280	0.0003	0.0	0.0
0.0001	50.0	300	0.0002	0.0	0.0
0.0001	53.3333	320	0.0002	0.0	0.0
0.0001	56.6667	340	0.0002	0.0	0.0
0.0001	60.0	360	0.0002	0.0	0.0
0.0001	63.3333	380	0.0002	0.0	0.0
0.0001	66.6667	400	0.0002	0.0	0.0
0.0001	70.0	420	0.0002	0.0	0.0
0.0000	73.3333	440	0.0002	0.0	0.0
0.0000	76.6667	460	0.0002	0.0	0.0
0.0000	80.0	480	0.0002	0.0	0.0
0.0000	83.3333	500	0.0002	0.0	0.0