whisper-tiny-ga-IE

This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4171
• Wer: 41.6175

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
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.1212	1.0	1586	0.7721	59.7573
0.7512	2.0	3172	0.5721	50.4880
0.5476	3.0	4758	0.4855	45.7250
0.4115	4.0	6344	0.4463	44.4064
0.315	5.0	7930	0.4237	43.0046
0.2525	6.0	9516	0.4170	41.9502
0.1916	7.0	11102	0.4120	41.7399
0.1596	8.0	12688	0.4129	41.3484
0.1322	9.0	14274	0.4158	41.1992
0.1158	10.0	15860	0.4171	41.6175

Framework versions

• Transformers 4.48.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0