whisper-large-v3-turbo-CAENNAIS_GB

This model is a fine-tuned version of openai/whisper-large-v3-turbo on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5002
• Wer: 18.0151

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: 32
• eval_batch_size: 16
• 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
No log	1.0	94	0.4424	23.3802
No log	2.0	188	0.3881	22.4889
No log	3.0	282	0.4092	19.4721
No log	4.0	376	0.4225	21.8375
No log	5.0	470	0.4463	21.7347
0.3774	6.0	564	0.4691	19.9349
0.3774	7.0	658	0.4626	23.0374
0.3774	8.0	752	0.4818	19.2664
0.3774	9.0	846	0.5083	18.6664
0.3774	10.0	940	0.5002	18.0151

Framework versions

• Transformers 4.46.2
• Pytorch 2.4.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.0