This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/common_voice_17_0 eu dataset. It achieves the following results on the evaluation set:

Loss: 0.3438
Wer: 9.3325

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: 64
eval_batch_size: 32
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: 40000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0781	2.3474	1000	0.2094	13.2938
0.03	4.6948	2000	0.2140	11.9132
0.0145	7.0423	3000	0.2377	11.5541
0.0086	9.3897	4000	0.2430	11.2078
0.0078	11.7371	5000	0.2533	11.2280
0.0061	14.0845	6000	0.2634	10.8927
0.0043	16.4319	7000	0.2667	10.6600
0.0053	18.7793	8000	0.2694	11.0154
0.0027	21.1268	9000	0.2741	10.7104
0.004	23.4742	10000	0.2827	11.2225
0.0041	25.8216	11000	0.2796	10.7232
0.0025	28.1690	12000	0.2887	10.7232
0.0032	30.5164	13000	0.2907	10.6820
0.0026	32.8638	14000	0.2950	10.9202
0.0016	35.2113	15000	0.2947	10.7672
0.0012	37.5587	16000	0.2980	10.4694
0.0015	39.9061	17000	0.2999	10.7030
0.0018	42.2535	18000	0.2982	10.8844
0.0008	44.6009	19000	0.3052	10.5638
0.002	46.9484	20000	0.3007	10.6966
0.0012	49.2958	21000	0.3069	10.3659
0.0017	51.6432	22000	0.3111	10.6655
0.0004	53.9906	23000	0.3161	10.0324
0.0006	56.3380	24000	0.3207	10.2340
0.0005	58.6854	25000	0.3111	10.1176
0.0003	61.0329	26000	0.3168	10.0333
0.0002	63.3803	27000	0.3247	10.0159
0.001	65.7277	28000	0.3191	10.3219
0.0001	68.0751	29000	0.3196	10.0617
0.0005	70.4225	30000	0.3221	10.1726
0.0001	72.7700	31000	0.3189	10.2138
0.0	75.1174	32000	0.3214	9.7759
0.0	77.4648	33000	0.3255	9.6348
0.0	79.8122	34000	0.3290	9.5432
0.0	82.1596	35000	0.3322	9.4864
0.0	84.5070	36000	0.3355	9.4525
0.0	86.8545	37000	0.3383	9.4168
0.0	89.2019	38000	0.3409	9.3921
0.0	91.5493	39000	0.3428	9.3481
0.0	93.8967	40000	0.3438	9.3325

Framework versions

Transformers 4.52.3
Pytorch 2.6.0+cu124
Datasets 3.6.0
Tokenizers 0.21.1

zuazo/whisper-small-eu-cv17_0

Whisper Small Basque

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

zuazo/
whisper-small-eu-cv17_0