zuazo/whisper-base-eu-cv22.0

openai/whisper-base

This model is a fine-tuned version of openai/whisper-base on the common_voice_22_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5965
• Wer: 14.1271

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: 3.75e-05
• train_batch_size: 128
• eval_batch_size: 64
• 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
• training_steps: 100000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0041	21.6450	5000	0.3770	16.8402
0.0025	43.2900	10000	0.4045	16.4480
0.0034	64.9351	15000	0.4193	16.2587
0.0014	86.5801	20000	0.4449	15.8953
0.0012	108.2251	25000	0.4490	15.8251
0.0	129.8701	30000	0.4585	14.3882
0.0	151.5152	35000	0.4865	13.9656
0.0	173.1602	40000	0.5256	13.7763
0.0	194.8052	45000	0.5518	13.6419
0.0	216.4502	50000	0.5558	13.6461
0.0	238.0952	55000	0.5683	13.7171
0.0	259.7403	60000	0.5801	13.9259
0.0	281.3853	65000	0.5886	13.9631
0.0	303.0303	70000	0.5965	14.1271

Framework versions

• Transformers 4.52.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1