You can deploy this model and start earning money today!

alinerodrigues/
wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-clean-10

Name: wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-clean-10
Availability: InStock

transformers

pytorch

wav2vec2

automatic-speech-recognition

generated_from_trainer

apache-2.0

endpoints_compatible

Views

No views yet

wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-clean-10

This model is a fine-tuned version of Edresson/wav2vec2-large-xlsr-coraa-portuguese on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.1766
Wer: 0.0913
Cer: 0.0291

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: 3e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 100
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
21.8058	1.0	67	8.5869	0.9997	0.9859
12.0549	2.0	134	5.9897	1.0	0.9719
5.5651	3.0	201	3.0869	1.0	1.0
5.5651	4.0	268	2.9406	1.0	1.0
2.9913	5.0	335	2.9153	1.0	1.0
2.9244	6.0	402	2.8824	1.0	1.0
2.9244	7.0	469	2.8784	1.0	1.0
2.8927	8.0	536	2.8759	1.0	1.0
2.8858	9.0	603	2.8599	1.0	1.0
2.8858	10.0	670	2.8328	1.0	1.0
2.8512	11.0	737	2.5475	1.0	0.9997
2.515	12.0	804	1.5878	1.0	0.5086
2.515	13.0	871	0.7009	0.8586	0.1889
1.3219	14.0	938	0.4050	0.2310	0.0598
0.6694	15.0	1005	0.3246	0.1755	0.0487
0.6694	16.0	1072	0.2909	0.1562	0.0439
0.4976	17.0	1139	0.2679	0.1420	0.0415
0.4105	18.0	1206	0.2519	0.1292	0.0384
0.4105	19.0	1273	0.2421	0.1194	0.0366
0.3865	20.0	1340	0.2325	0.1184	0.0354
0.317	21.0	1407	0.2251	0.1096	0.0341
0.317	22.0	1474	0.2195	0.1092	0.0339
0.2925	23.0	1541	0.2106	0.1018	0.0320
0.2721	24.0	1608	0.2072	0.0981	0.0317
0.2721	25.0	1675	0.2106	0.0981	0.0317
0.2531	26.0	1742	0.2046	0.1042	0.0330
0.2634	27.0	1809	0.2071	0.1001	0.0321
0.2634	28.0	1876	0.2028	0.1042	0.0328
0.2391	29.0	1943	0.1973	0.0957	0.0308
0.2232	30.0	2010	0.2017	0.0974	0.0313
0.2232	31.0	2077	0.1987	0.0974	0.0308
0.2111	32.0	2144	0.1898	0.0920	0.0298
0.2121	33.0	2211	0.2006	0.0954	0.0314
0.2121	34.0	2278	0.1934	0.0920	0.0303
0.1868	35.0	2345	0.1921	0.0944	0.0306
0.1869	36.0	2412	0.1884	0.0893	0.0292
0.1869	37.0	2479	0.1903	0.0866	0.0292
0.1935	38.0	2546	0.1867	0.0900	0.0294
0.1957	39.0	2613	0.1874	0.0927	0.0298
0.1957	40.0	2680	0.1845	0.0923	0.0297
0.1772	41.0	2747	0.1862	0.0927	0.0298
0.1748	42.0	2814	0.1894	0.0906	0.0297
0.1748	43.0	2881	0.1816	0.0933	0.0301
0.1498	44.0	2948	0.1795	0.0920	0.0296
0.1606	45.0	3015	0.1867	0.0906	0.0299
0.1606	46.0	3082	0.1866	0.0886	0.0294
0.1599	47.0	3149	0.1883	0.0920	0.0300
0.1487	48.0	3216	0.1802	0.0933	0.0298
0.1487	49.0	3283	0.1808	0.0937	0.0298
0.148	50.0	3350	0.1824	0.0916	0.0292
0.1457	51.0	3417	0.1843	0.0893	0.0293
0.1457	52.0	3484	0.1822	0.0923	0.0293
0.1472	53.0	3551	0.1766	0.0913	0.0291
0.1413	54.0	3618	0.1811	0.0933	0.0292
0.1413	55.0	3685	0.1807	0.0906	0.0291
0.1357	56.0	3752	0.1808	0.0879	0.0284
0.1382	57.0	3819	0.1810	0.0933	0.0296
0.1382	58.0	3886	0.1817	0.0910	0.0287
0.1371	59.0	3953	0.1844	0.0889	0.0286
0.141	60.0	4020	0.1883	0.0883	0.0284
0.141	61.0	4087	0.1864	0.0930	0.0290
0.147	62.0	4154	0.1861	0.0920	0.0289
0.1316	63.0	4221	0.1863	0.0950	0.0296
0.1316	64.0	4288	0.1909	0.0950	0.0302
0.1329	65.0	4355	0.1880	0.0913	0.0291
0.1326	66.0	4422	0.1851	0.0930	0.0291
0.1326	67.0	4489	0.1842	0.0937	0.0292
0.1345	68.0	4556	0.1856	0.0957	0.0297
0.1371	69.0	4623	0.1840	0.0927	0.0291
0.1371	70.0	4690	0.1845	0.0923	0.0292
0.1325	71.0	4757	0.1806	0.0920	0.0288
0.1264	72.0	4824	0.1810	0.0923	0.0289
0.1264	73.0	4891	0.1836	0.0944	0.0295

Framework versions

Transformers 4.28.0
Pytorch 2.2.1+cu121
Datasets 2.17.0
Tokenizers 0.13.3

alinerodrigues/wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-clean-10

wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-clean-10

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

alinerodrigues/
wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-clean-10