This model is a fine-tuned version of openai/whisper-small on the Common Voice 13 - whisper_small on ASR_JA dataset. It achieves the following results on the evaluation set:

Loss: 0.5735
Wer Ortho: 731.1720
Wer: 622.0128

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: 1e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant_with_warmup
lr_scheduler_warmup_steps: 50
training_steps: 4000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.1701	1.13	500	0.4424	480.6484	2104.7685
0.0606	2.26	1000	0.4455	238.1393	1241.4749
0.0334	3.39	1500	0.4821	496.1941	1038.0787
0.0167	4.52	2000	0.4851	614.7604	1104.6160
0.0113	5.66	2500	0.5100	265.9686	457.5270
0.007	6.79	3000	0.5387	338.3004	569.8087
0.0072	7.92	3500	0.5301	213.4515	597.5880
0.005	9.05	4000	0.5735	731.1720	622.0128

Framework versions

Transformers 4.37.2
Pytorch 2.1.0+cu121
Datasets 2.16.1
Tokenizers 0.15.1

Vickyee/whisper-small-ja

Whisper Small Ja - Vick

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Vickyee/
whisper-small-ja