Gummybear05/whisper-small-E50

whisper-small-E30

This model is a fine-tuned version of openai/whisper-small on the aihub old adult freq speed pause changed dataset. It achieves the following results on the evaluation set:

• Loss: 0.1691
• Cer: 4.5465

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use 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: 50
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.7214	0.2581	100	0.2507	5.9680
0.3927	0.5161	200	0.1969	5.3924
0.357	0.7742	300	0.1870	5.2044
0.2509	1.0310	400	0.1762	4.8520
0.1611	1.2890	500	0.1752	4.6346
0.159	1.5471	600	0.1713	4.5172
0.1578	1.8052	700	0.1691	4.5465

Framework versions

• Transformers 4.48.0.dev0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0