whisper-large-v3-CAENNAIS-DI

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

• Loss: 0.5385
• Wer: 22.8487

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: 2e-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: 150
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	51	0.3782	19.2935
No log	2.0	102	0.4309	24.9547
No log	3.0	153	0.4866	27.1060
No log	4.0	204	0.5200	25.1359
No log	5.0	255	0.5453	25.9737
No log	6.0	306	0.5414	25.2717
No log	7.0	357	0.5434	25.7926
No log	8.0	408	0.5569	26.5625
No log	9.0	459	0.5425	27.6042
0.1412	10.0	510	0.5385	22.8487

Framework versions

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