evanarlian/whisper-base-id

Whisper Base Indonesian

This model is a fine-tuned version of openai/whisper-base on the mozilla-foundation/common_voice_11_0 id dataset. It achieves the following results on the evaluation set:

• Loss: 0.5101
• Wer: 23.7573

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: 2e-05
• train_batch_size: 64
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2041	4.95	500	0.3906	23.9140
0.015	9.9	1000	0.4619	24.3752
0.0032	14.85	1500	0.4901	23.7803
0.0015	19.8	2000	0.5101	23.7573
0.001	24.75	2500	0.5265	23.9786
0.0008	29.7	3000	0.5399	24.1216
0.0006	34.65	3500	0.5501	23.8956
0.0005	39.6	4000	0.5583	24.0570
0.0004	44.55	4500	0.5638	24.1815
0.0004	49.5	5000	0.5659	24.1492

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.0+cu117
• Datasets 2.7.1.dev0
• Tokenizers 0.13.2