infinite-learning-station/whisper-tiny-indian-restaurant

Whisper Tiny Indian Restaurant

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

• Loss: 0.0593
• Wer: 12.9686

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_FUSED 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
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0954	0.9893	1200	0.0963	17.5894
0.0765	1.9786	2400	0.0794	16.6739
0.0658	2.9678	3600	0.0725	15.8521
0.0627	3.9571	4800	0.0665	15.2682
0.0505	4.9464	6000	0.0648	14.4968
0.0471	5.9357	7200	0.0613	13.9129
0.0404	6.9250	8400	0.0600	13.3218
0.0323	7.9143	9600	0.0600	12.9181
0.0289	8.9035	10800	0.0592	12.7235
0.0251	9.8928	12000	0.0593	12.9686

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.0+cu126
• Datasets 4.4.2
• Tokenizers 0.22.1