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: 332
training_steps: 4740

Training Loss	Epoch	Step	Validation Loss	Wer	Iu Precision	Iu Recall	Iu F1	Iu Tp	Iu Fp	Iu Fn
1.0674	0.4998	592	0.5207	0.2533	0.7185	0.6823	0.6999	393	154	183
0.9350	0.9996	1184	0.4790	0.2468	0.6972	0.8073	0.7482	465	202	111
0.5897	1.4989	1776	0.4782	0.2251	0.8164	0.6406	0.7179	369	83	207
0.5256	1.9987	2368	0.4634	0.2371	0.7393	0.8125	0.7742	468	165	108
0.2874	2.4981	2960	0.4950	0.2204	0.7442	0.7778	0.7606	448	154	128
0.2867	2.9979	3552	0.4903	0.2194	0.7559	0.7795	0.7675	449	145	127
0.0946	3.4973	4144	0.5551	0.2141	0.7810	0.7552	0.7679	435	122	141
0.0899	3.9970	4736	0.5486	0.2120	0.7742	0.7917	0.7828	456	133	120
0.0899	4.0	4740	0.5486	0.2119	0.7742	0.7917	0.7828	456	133	120

nilc-nlp/psst-portuguese-4e-1s-difflib

psst-portuguese-4e-1s-difflib