This model is a fine-tuned version of openai/whisper-large-v3 on the OUTCOMESAI/medical_n_common_speech_corpus_50_50 en dataset. It achieves the following results on the evaluation set:

Loss: 0.3196
Wer: 5.2186

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e-07
train_batch_size: 64
eval_batch_size: 32
seed: 42
distributed_type: multi-GPU
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: 250
training_steps: 5000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
5.126	0.0969	250	0.3694	5.6601
4.367	0.1938	500	0.3586	5.8156
4.1514	0.2907	750	0.3511	5.8839
3.962	0.3876	1000	0.3450	5.7805
3.9038	0.4845	1250	0.3403	6.1746
3.8313	0.5814	1500	0.3359	5.9738
3.7778	0.6783	1750	0.3333	5.9218
3.7421	0.7752	2000	0.3306	6.1327
3.7367	0.8721	2250	0.3281	5.6561
3.6878	0.9690	2500	0.3257	5.5154
3.6769	1.0659	2750	0.3242	5.4803
3.6508	1.1628	3000	0.3235	5.4634
3.6292	1.2597	3250	0.3220	5.3512
3.6179	1.3566	3500	0.3210	5.2254
3.6032	1.4535	3750	0.3206	5.2207
3.5922	1.5504	4000	0.3201	5.3038
3.5743	1.6473	4250	0.3198	5.2633
3.5882	1.7442	4500	0.3198	5.2254
3.6021	1.8411	4750	0.3196	5.2186
3.5865	1.9380	5000	0.3193	5.2213

Framework versions

Transformers 4.48.0.dev0
Pytorch 2.1.0+cu118
Datasets 3.2.1.dev0
Tokenizers 0.21.0

saurabhy27-outcomes/whisper-large-v3-common-n-medical-50-50

Whisper Large V3 Common n Medical 50 50

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

saurabhy27-outcomes/
whisper-large-v3-common-n-medical-50-50