erjoy/whisper-small-hi-v1

Whisper tiny Hi - v1

This model is a fine-tuned version of openai/whisper-tiny on the Common Voice 17.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5045
• Wer: 56.2682

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: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 2500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4841	1.1236	500	0.6430	68.6439
0.3069	2.2472	1000	0.5336	61.4432
0.2537	3.3708	1500	0.5145	58.1104
0.2111	4.4944	2000	0.5054	56.4325
0.191	5.6180	2500	0.5045	56.2682

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.0