zuazo/whisper-large-v3-turbo-eu-cv17_0

Whisper Large-V3-Turbo Basque

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the mozilla-foundation/common_voice_17_0 eu dataset. It achieves the following results on the evaluation set:

• Loss: 0.2907
• Wer: 7.3546

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: 32
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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
0.1123	2.3474	1000	0.2163	13.2370
0.067	4.6948	2000	0.2017	11.0319
0.0327	7.0423	3000	0.2156	10.7736
0.0243	9.3897	4000	0.2262	10.4190
0.0198	11.7371	5000	0.2324	10.5500
0.0129	14.0845	6000	0.2427	10.0645
0.0133	16.4319	7000	0.2425	9.9225
0.0123	18.7793	8000	0.2540	10.3851
0.0085	21.1268	9000	0.2441	9.5423
0.0091	23.4742	10000	0.2557	9.7585
0.0076	25.8216	11000	0.2545	9.6660
0.0062	28.1690	12000	0.2590	9.5139
0.0044	30.5164	13000	0.2605	9.2290
0.0046	32.8638	14000	0.2680	9.5588
0.0057	35.2113	15000	0.2734	9.8859
0.0041	37.5587	16000	0.2703	9.3087
0.0057	39.9061	17000	0.2607	9.1566
0.002	42.2535	18000	0.2685	8.8516
0.0034	44.6009	19000	0.2780	9.5964
0.0023	46.9484	20000	0.2801	9.3298
0.0022	49.2958	21000	0.2728	9.0174
0.0015	51.6432	22000	0.2747	8.9532
0.001	53.9906	23000	0.2745	8.4649
0.0018	56.3380	24000	0.2807	9.1026
0.0013	58.6854	25000	0.2686	8.4915
0.0009	61.0329	26000	0.2767	8.3010
0.0012	63.3803	27000	0.2850	8.8241
0.0012	65.7277	28000	0.2830	8.6573
0.0002	68.0751	29000	0.2806	8.3165
0.0007	70.4225	30000	0.2802	8.5419
0.0006	72.7700	31000	0.2823	8.3532
0.0002	75.1174	32000	0.2774	8.2231
0.0003	77.4648	33000	0.2807	8.3120
0.0001	79.8122	34000	0.2834	7.9043
0.0	82.1596	35000	0.2821	7.7192
0.0	84.5070	36000	0.2841	7.5442
0.0	86.8545	37000	0.2865	7.4645
0.0	89.2019	38000	0.2885	7.4306
0.0	91.5493	39000	0.2900	7.3665
0.0	93.8967	40000	0.2907	7.3546

Framework versions

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