whisper-large-v3-turbo-half-stage-2

This model is a fine-tuned version of JacobLinCool/whisper-large-v3-turbo-half on the common_voice_16_1 dataset. It achieves the following results on the evaluation set:

Loss: 0.8327
Wer: 30.7358

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: 0.0002
train_batch_size: 16
eval_batch_size: 32
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 5000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0	0	0.7508	25.6132
1.1243	0.1	500	1.5399	75.5155
0.8571	0.2	1000	1.3668	61.1895
0.7222	0.3	1500	1.2842	70.1324
0.6757	0.4	2000	1.1646	50.5101
0.5335	0.5	2500	1.0503	40.9811
0.5068	0.6	3000	0.9836	36.8135
0.4505	0.7	3500	0.9124	33.7964
0.4378	0.8	4000	0.8649	36.5965
0.4292	0.9	4500	0.8402	37.7035
0.3966	1.0	5000	0.8327	30.7358