whisper-medium-bigcgen-female-5hrs-62

This model is a fine-tuned version of openai/whisper-medium on the bigcgen dataset. It achieves the following results on the evaluation set:

• Loss: 0.7278
• Wer: 0.5489

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: 2
• eval_batch_size: 2
• seed: 62
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.0539	0.5984	200	1.0387	0.6708
0.7105	1.1945	400	0.8432	0.5764
0.6344	1.7928	600	0.7586	0.5600
0.5221	2.3889	800	0.7457	0.5623
0.4151	2.9873	1000	0.7278	0.5489
0.219	3.5834	1200	0.7980	0.5287
0.1226	4.1795	1400	0.8193	0.5362
0.1245	4.7779	1600	0.8030	0.4976
0.0529	5.3740	1800	0.8765	0.4963

Framework versions

• Transformers 4.53.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.0