learning_rate: 1e-05
train_batch_size: 4
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 8
optimizer: Use 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: 664
training_steps: 9480

Training Loss	Epoch	Step	Validation Loss	Wer	Iu F1	Iu Tp	Iu Fp	Iu Fn
1.0142	1.0	1185	0.5133	0.2418	0.7429	468	216	108
0.6599	2.0	2370	0.5071	0.2629	0.7538	427	130	149
0.3495	3.0	3555	0.5182	0.2406	0.7448	464	206	112
0.1633	4.0	4740	0.5935	0.2408	0.7675	434	121	142
0.0656	5.0	5925	0.6209	0.2346	0.7527	455	178	121
0.0320	6.0	7110	0.7027	0.2306	0.7612	443	145	133
0.0118	7.0	8295	0.7577	0.2247	0.7552	421	118	155
0.0028	8.0	9480	0.7910	0.2235	0.7698	443	132	133

nilc-nlp/psst-model-8e-1s-lp3200hz

psst-model-8e-1s-lp3200hz