whisper-large-v3-turbo-CAENNAIS_GB_E1

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: 0.6754
• Wer: 27.0515

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	39	1.3360	65.4580
No log	2.0	78	0.5744	30.4389
No log	3.0	117	0.5146	26.1927
No log	4.0	156	0.5115	33.5878
No log	5.0	195	0.5258	25.0954
No log	6.0	234	0.5566	25.4294
No log	7.0	273	0.5865	25.8588
No log	8.0	312	0.6115	26.9084
No log	9.0	351	0.6361	25.6202
No log	10.0	390	0.6754	27.0515

Framework versions

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