hanasim/breeze-dsw-small-id

Breeze DSW Indonesian - small

This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/common_voice_16_0 id dataset. It achieves the following results on the evaluation set:

• Loss: 0.3159
• Wer: 17.7063

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: 32
• eval_batch_size: 16
• seed: 42
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4199	0.1	100	0.3494	19.1600
0.282	1.04	200	0.3159	17.7063
0.1241	1.14	300	0.3291	18.5988
0.1232	2.09	400	0.3279	18.8886
0.064	3.03	500	0.3389	17.9133
0.0305	3.13	600	0.3557	18.4792
0.0282	4.08	700	0.3625	18.0559
0.0117	5.02	800	0.3699	18.2906
0.0079	5.12	900	0.3794	18.3596
0.0081	6.06	1000	0.3826	18.2906

Framework versions

• Transformers 4.37.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.16.2.dev0
• Tokenizers 0.15.0