learning_rate: 1e-05
train_batch_size: 64
eval_batch_size: 64
seed: 42
optimizer: Use 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: 0.04
training_steps: 46000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Raw	Cer Raw	Wer	Cer
0.2105	0.0217	1000	0.4291	21.0947	7.3489	21.0123	7.3330
0.2114	0.0435	2000	0.3882	20.2421	7.6348	20.1920	7.6252
0.1814	0.0652	3000	0.3561	17.8716	6.2992	17.8494	6.2949
0.1671	0.0870	4000	0.3379	17.5020	6.2904	17.4811	6.2869
0.1378	0.1087	5000	0.3265	17.1660	6.2993	17.1565	6.2975
0.1211	0.1304	6000	0.3199	17.3816	6.5366	17.3708	6.5346
0.1371	0.1522	7000	0.3200	16.8078	6.1243	16.8034	6.1235
0.1474	0.1739	8000	0.2990	15.9723	6.0495	15.9704	6.0491
0.1405	0.1957	9000	0.2938	15.5342	5.6053	15.5342	5.6053
0.2387	0.2174	10000	0.3001	15.7516	5.7251	15.7504	5.7248
0.1441	0.2391	11000	0.2876	14.7728	5.3077	14.7722	5.3076
0.1232	0.2609	12000	0.2803	14.7005	5.2640	14.7005	5.2640
0.2715	0.2826	13000	0.2770	14.7481	5.3805	14.7468	5.3803
0.2600	0.3043	14000	0.2833	15.4372	5.8523	15.4366	5.8522
0.1616	0.3261	15000	0.2715	14.2402	5.0899	14.2396	5.0898
0.2634	0.3478	16000	0.2715	14.1122	5.0557	14.1109	5.0555
0.2724	0.3696	17000	0.2643	14.2193	5.2369	14.2187	5.2368
0.1641	0.3913	18000	0.2551	13.6678	5.0883	13.6678	5.0883
0.1795	0.4130	19000	0.2547	14.0418	5.3390	14.0418	5.3390
0.2134	0.4348	20000	0.2497	13.8301	5.3565	13.8301	5.3565
0.2445	0.4565	21000	0.2481	13.7248	5.1595	13.7248	5.1595
0.3013	0.4783	22000	0.2461	13.4941	4.9600	13.4941	4.9600
0.2151	0.5	23000	0.2433	13.1035	4.8197	13.1035	4.8197
0.1280	1.0040	24000	0.2368	13.4332	5.1921	13.4332	5.1921
0.0863	1.0257	25000	0.2337	12.6167	4.7145	12.6167	4.7145
0.0799	1.0474	26000	0.2339	12.8316	4.8257	12.8316	4.8257
0.1029	1.0692	27000	0.2329	12.6167	4.5999	12.6167	4.5999
0.0867	1.0909	28000	0.2330	12.7244	4.7077	12.7244	4.7077
0.0840	1.1127	29000	0.2310	12.5786	4.6454	12.5786	4.6454
0.1024	1.1344	30000	0.2316	12.3517	4.6403	12.3517	4.6403
0.0929	1.1561	31000	0.2297	12.5044	4.7394	12.5044	4.7394
0.1030	1.1779	32000	0.2283	12.3574	4.6086	12.3574	4.6086
0.1092	1.1996	33000	0.2310	12.3542	4.6186	12.3542	4.6186
0.0939	1.2213	34000	0.2308	12.6959	4.8683	12.6959	4.8683
0.0722	1.2431	35000	0.2283	12.4309	4.7734	12.4309	4.7734
0.0916	1.2648	36000	0.2283	12.3269	4.6052	12.3269	4.6052
0.1084	1.2866	37000	0.2291	12.6610	4.8985	12.6610	4.8985
0.1796	1.3083	38000	0.2300	12.3288	4.5390	12.3288	4.5390
0.1343	1.3300	39000	0.2262	12.3853	4.6860	12.3853	4.6860
0.1363	1.3518	40000	0.2268	12.0664	4.4976	12.0664	4.4976
0.1393	1.3735	41000	0.2237	12.2699	4.7344	12.2699	4.7344
0.1617	1.3953	42000	0.2231	12.4189	4.7940	12.4189	4.7940
0.2268	1.417	43000	0.2240	12.0207	4.4868	12.0207	4.4868
0.3400	1.4387	44000	0.2238	12.2071	4.5810	12.2071	4.5810
0.1902	1.4605	45000	0.2238	12.0809	4.5142	12.0809	4.5142
0.2122	1.4822	46000	0.2225	12.4062	4.8335	12.4062	4.8335

Framework versions

Transformers 5.14.1
Pytorch 2.6.0+cu124
Datasets 5.0.1
Tokenizers 0.22.2

Citation

Please cite the model using the following BibTeX entry:

bibtex

@misc{deepdml/whisper-base-es-mix-norm,
      title={Fine-tuned Whisper base ASR model for speech recognition in Spanish},
      author={Jimenez, David},
      howpublished={\url{https://huggingface.co/deepdml/whisper-base-es-mix-norm}},
      year={2026}
    }