evanarlian/whisper-tiny-id

Whisper Tiny Indonesian

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

• Loss: 0.6202
• Wer: 32.4218

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
0.3823	4.95	500	0.5251	33.4732
0.0495	9.9	1000	0.5700	33.3902
0.0077	14.85	1500	0.6202	32.4218
0.0031	19.8	2000	0.6616	32.5371
0.0019	24.75	2500	0.6873	32.7954
0.0014	29.7	3000	0.7056	33.5700
0.0011	34.65	3500	0.7204	33.7960
0.0009	39.6	4000	0.7327	33.7729
0.0008	44.55	4500	0.7400	33.9113
0.0007	49.5	5000	0.7428	33.3441

Framework versions

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