You can deploy this model and start earning money today!

alinerodrigues/
wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-02

Name: wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-02
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-02

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: 0.1708
Wer: 0.0912
Cer: 0.0295

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
27.7048	1.0	86	3.1863	1.0	1.0
7.6657	2.0	172	2.9513	1.0	1.0
3.0186	3.0	258	2.9043	1.0	1.0
2.9273	4.0	344	2.8893	1.0	1.0
2.8917	5.0	430	2.3987	0.9968	0.9289
2.1369	6.0	516	0.7196	0.3304	0.0875
0.9711	7.0	602	0.4605	0.2265	0.0617
0.9711	8.0	688	0.3820	0.1878	0.0543
0.6721	9.0	774	0.3183	0.1714	0.0494
0.5383	10.0	860	0.2943	0.1573	0.0470
0.4648	11.0	946	0.2775	0.1417	0.0432
0.4224	12.0	1032	0.2601	0.1407	0.0424
0.4083	13.0	1118	0.2475	0.1229	0.0385
0.339	14.0	1204	0.2397	0.1175	0.0378
0.339	15.0	1290	0.2227	0.1085	0.0356
0.3152	16.0	1376	0.2207	0.1080	0.0348
0.2875	17.0	1462	0.2112	0.1075	0.0352
0.2808	18.0	1548	0.2027	0.1058	0.0342
0.2788	19.0	1634	0.2034	0.1012	0.0336
0.2514	20.0	1720	0.1978	0.1027	0.0333
0.2449	21.0	1806	0.1933	0.0997	0.0329
0.2449	22.0	1892	0.1948	0.0985	0.0320
0.2556	23.0	1978	0.2003	0.0973	0.0327
0.2426	24.0	2064	0.1917	0.0992	0.0328
0.2357	25.0	2150	0.1837	0.1000	0.0326
0.2205	26.0	2236	0.1812	0.1005	0.0324
0.1978	27.0	2322	0.1869	0.0961	0.0320
0.2175	28.0	2408	0.1864	0.0946	0.0315
0.2175	29.0	2494	0.1895	0.0975	0.0315
0.2042	30.0	2580	0.1824	0.0992	0.0317
0.206	31.0	2666	0.1869	0.0956	0.0307
0.1929	32.0	2752	0.1847	0.0931	0.0306
0.1709	33.0	2838	0.1788	0.0934	0.0301
0.1851	34.0	2924	0.1795	0.0917	0.0299
0.1658	35.0	3010	0.1764	0.0944	0.0300
0.1658	36.0	3096	0.1767	0.0941	0.0293
0.1738	37.0	3182	0.1739	0.0961	0.0303
0.1691	38.0	3268	0.1795	0.0900	0.0295
0.1735	39.0	3354	0.1791	0.0902	0.0295
0.1913	40.0	3440	0.1804	0.0902	0.0302
0.1645	41.0	3526	0.1779	0.0914	0.0293
0.1544	42.0	3612	0.1725	0.0934	0.0294
0.1544	43.0	3698	0.1735	0.0900	0.0289
0.1673	44.0	3784	0.1708	0.0912	0.0295
0.1672	45.0	3870	0.1709	0.0902	0.0289
0.1729	46.0	3956	0.1746	0.0880	0.0292
0.1483	47.0	4042	0.1738	0.0902	0.0284
0.1571	48.0	4128	0.1771	0.0900	0.0291
0.1408	49.0	4214	0.1806	0.0907	0.0294
0.1358	50.0	4300	0.1777	0.0892	0.0293
0.1358	51.0	4386	0.1781	0.0892	0.0289
0.1662	52.0	4472	0.1777	0.0868	0.0283
0.1329	53.0	4558	0.1768	0.0871	0.0286
0.1341	54.0	4644	0.1764	0.0905	0.0290
0.1319	55.0	4730	0.1757	0.0892	0.0293
0.1377	56.0	4816	0.1780	0.0868	0.0277
0.1211	57.0	4902	0.1828	0.0910	0.0290
0.1211	58.0	4988	0.1821	0.0875	0.0289
0.1436	59.0	5074	0.1801	0.0902	0.0292
0.1395	60.0	5160	0.1751	0.0871	0.0281
0.1276	61.0	5246	0.1777	0.0885	0.0288
0.131	62.0	5332	0.1788	0.0866	0.0282
0.1276	63.0	5418	0.1810	0.0853	0.0284
0.1205	64.0	5504	0.1819	0.0856	0.0284

Framework versions

Transformers 4.28.0
Pytorch 2.1.1+cu121
Datasets 2.15.0
Tokenizers 0.13.3

alinerodrigues/wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-02

wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-02

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-02