learning_rate: 1e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
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: 500
training_steps: 20000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
0.1896	4.9751	1000	0.1947	49.0934
0.1242	9.9502	2000	0.1866	45.1973
0.0926	14.9254	3000	0.1925	44.0456
0.0668	19.9005	4000	0.2036	44.3396
0.0461	24.8756	5000	0.2099	43.4207
0.0338	29.8507	6000	0.2164	42.1588
0.0256	34.8259	7000	0.2272	42.5018
0.0245	39.8010	8000	0.2263	42.1588
0.019	44.7761	9000	0.2368	41.1786
0.0137	49.7512	10000	0.2405	41.4359
0.0122	54.7264	11000	0.2425	41.6074
0.0115	59.7015	12000	0.2483	41.8402
0.0081	64.6766	13000	0.2506	40.8968
0.01	69.6517	14000	0.2561	41.4359
0.0071	74.6269	15000	0.2592	41.8525
0.0063	79.6020	16000	0.2576	40.8846
0.0053	84.5771	17000	0.2614	40.8601
0.0052	89.5522	18000	0.2594	40.9091
0.0049	94.5274	19000	0.2616	41.3747
0.005	99.5025	20000	0.2619	41.1664

surafelabebe/whisper-small-am-v1

Whisper Small Amharic