whisper-small-fine-tuned-with-patient-conversations

This model is a fine-tuned version of openai/whisper-small on the audiofolder dataset. It achieves the following results on the evaluation set:

Model description

More information needed

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 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: 1000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
3.4179	1.0526	40	0.9737	1.4178	1.4518
0.8089	2.1053	80	0.5369	1.3510	1.2829
0.4491	3.1579	120	0.2198	0.1097	0.0472
0.093	4.2105	160	0.2152	0.1092	0.0525
0.0381	5.2632	200	0.2281	0.0941	0.0491
0.0226	6.3158	240	0.2255	0.0936	0.0434
0.0153	7.3684	280	0.2422	0.0946	0.0468
0.0057	8.4211	320	0.2523	0.1014	0.0539
0.0044	9.4737	360	0.2679	0.0936	0.0510
0.0031	10.5263	400	0.2646	0.0922	0.0520
0.0014	11.5789	440	0.2766	0.1297	0.0816
0.0008	12.6316	480	0.2782	0.1273	0.0811
0.0004	13.6842	520	0.2796	0.1336	0.0887
0.0003	14.7368	560	0.2825	0.1302	0.0854
0.0002	15.7895	600	0.2846	0.1312	0.0864
0.0002	16.8421	640	0.2877	0.1331	0.0873
0.0002	17.8947	680	0.2898	0.1346	0.0883
0.0001	18.9474	720	0.2917	0.1346	0.0883
0.0001	20.0	760	0.2937	0.1351	0.0883
0.0001	21.0526	800	0.2957	0.1370	0.0911
0.0001	22.1053	840	0.2972	0.1370	0.0911
0.0001	23.1579	880	0.2989	0.1370	0.0911
0.0001	24.2105	920	0.3001	0.1370	0.0911
0.0001	25.2632	960	0.3014	0.1438	0.0978
0.0001	26.3158	1000	0.3028	0.1433	0.0964