simpragma/breeze-listen-dsw-base-id

Breeze DSW Indonesian - base

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

• Loss: 0.5347
• Wer: 32.8273

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: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5452	1.02	200	0.5464	35.1688
0.3445	2.04	400	0.5405	34.0694
0.1397	3.07	600	0.5347	32.8273
0.0988	5.01	800	0.5654	35.6749
0.077	6.03	1000	0.5786	33.9452
0.0338	7.05	1200	0.6050	33.9820
0.0137	8.08	1400	0.6221	34.1016
0.0153	10.02	1600	0.6431	33.9038
0.0125	11.04	1800	0.6514	33.7520
0.0092	12.06	2000	0.6528	33.8256

Framework versions

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