wav2vec2-large-xlsr-coraa-exp-12

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: 0.5650
• Wer: 0.3527
• Cer: 0.1823

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: 150
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
37.6216	1.0	14	23.2071	1.0	0.9619
37.6216	2.0	28	6.9366	1.0	0.9619
37.6216	3.0	42	4.4250	1.0	0.9619
37.6216	4.0	56	3.9154	1.0	0.9619
37.6216	5.0	70	3.6849	1.0	0.9619
37.6216	6.0	84	3.5283	1.0	0.9619
37.6216	7.0	98	3.3716	1.0	0.9619
8.823	8.0	112	3.2657	1.0	0.9619
8.823	9.0	126	3.1796	1.0	0.9619
8.823	10.0	140	3.1568	1.0	0.9619
8.823	11.0	154	3.1071	1.0	0.9619
8.823	12.0	168	3.0891	1.0	0.9619
8.823	13.0	182	3.0588	1.0	0.9619
8.823	14.0	196	3.0422	1.0	0.9619
3.0574	15.0	210	3.0388	1.0	0.9619
3.0574	16.0	224	3.0324	1.0	0.9619
3.0574	17.0	238	3.0253	1.0	0.9619
3.0574	18.0	252	3.0100	1.0	0.9619
3.0574	19.0	266	3.0079	1.0	0.9619
3.0574	20.0	280	3.0150	1.0	0.9619
3.0574	21.0	294	3.0033	1.0	0.9619
2.95	22.0	308	2.9999	1.0	0.9619
2.95	23.0	322	2.9940	1.0	0.9619
2.95	24.0	336	2.9982	1.0	0.9619
2.95	25.0	350	3.0212	1.0	0.9619
2.95	26.0	364	2.9951	1.0	0.9619
2.95	27.0	378	2.9893	1.0	0.9619
2.95	28.0	392	2.9907	1.0	0.9619
2.9233	29.0	406	2.9889	1.0	0.9619
2.9233	30.0	420	2.9813	1.0	0.9619
2.9233	31.0	434	2.9795	1.0	0.9619
2.9233	32.0	448	2.9633	1.0	0.9619
2.9233	33.0	462	2.9653	1.0	0.9585
2.9233	34.0	476	2.9050	1.0	0.9619
2.9233	35.0	490	2.8806	1.0	0.9619
2.8852	36.0	504	2.8230	1.0	0.9619
2.8852	37.0	518	2.7805	1.0	0.9619
2.8852	38.0	532	2.7044	1.0	0.9572
2.8852	39.0	546	2.6561	1.0	0.9559
2.8852	40.0	560	2.5475	1.0	0.9254
2.8852	41.0	574	2.3336	1.0	0.7458
2.8852	42.0	588	2.0696	1.0	0.5468
2.5339	43.0	602	1.7760	1.0	0.4971
2.5339	44.0	616	1.5433	1.0	0.4546
2.5339	45.0	630	1.3529	1.0	0.4067
2.5339	46.0	644	1.2149	0.9998	0.3834
2.5339	47.0	658	1.0925	0.9943	0.3578
2.5339	48.0	672	1.0236	0.8954	0.3129
2.5339	49.0	686	0.9525	0.7062	0.2623
1.3395	50.0	700	0.8922	0.5063	0.2201
1.3395	51.0	714	0.8068	0.4774	0.2115
1.3395	52.0	728	0.7932	0.4553	0.2076
1.3395	53.0	742	0.7726	0.4453	0.2066
1.3395	54.0	756	0.7551	0.4340	0.2027
1.3395	55.0	770	0.7420	0.4305	0.2039
1.3395	56.0	784	0.7146	0.4212	0.2008
1.3395	57.0	798	0.6768	0.4096	0.1957
0.7419	58.0	812	0.6767	0.4080	0.1962
0.7419	59.0	826	0.6709	0.4069	0.1971
0.7419	60.0	840	0.6791	0.4025	0.1967
0.7419	61.0	854	0.6560	0.4029	0.1938
0.7419	62.0	868	0.6474	0.3976	0.1939
0.7419	63.0	882	0.6584	0.3982	0.1941
0.7419	64.0	896	0.6619	0.3960	0.1938
0.5254	65.0	910	0.6514	0.3923	0.1936
0.5254	66.0	924	0.6363	0.3874	0.1915
0.5254	67.0	938	0.6173	0.3797	0.1900
0.5254	68.0	952	0.6284	0.3887	0.1918
0.5254	69.0	966	0.6153	0.3767	0.1897
0.5254	70.0	980	0.6084	0.3736	0.1879
0.5254	71.0	994	0.6196	0.3773	0.1900
0.4219	72.0	1008	0.6075	0.3730	0.1899
0.4219	73.0	1022	0.6017	0.3712	0.1884
0.4219	74.0	1036	0.5947	0.3694	0.1872
0.4219	75.0	1050	0.5975	0.3696	0.1889
0.4219	76.0	1064	0.6020	0.3728	0.1887
0.4219	77.0	1078	0.5994	0.3704	0.1892
0.4219	78.0	1092	0.5822	0.3716	0.1877
0.385	79.0	1106	0.6073	0.3742	0.1893
0.385	80.0	1120	0.6029	0.3728	0.1874
0.385	81.0	1134	0.5961	0.3700	0.1868
0.385	82.0	1148	0.6032	0.3702	0.1870
0.385	83.0	1162	0.6115	0.3722	0.1889
0.385	84.0	1176	0.6018	0.3690	0.1883
0.385	85.0	1190	0.5824	0.3665	0.1855
0.3463	86.0	1204	0.5985	0.3669	0.1866
0.3463	87.0	1218	0.5833	0.3669	0.1861
0.3463	88.0	1232	0.5775	0.3637	0.1862
0.3463	89.0	1246	0.5747	0.3606	0.1850
0.3463	90.0	1260	0.5784	0.3639	0.1851
0.3463	91.0	1274	0.5841	0.3604	0.1858
0.3463	92.0	1288	0.5762	0.3655	0.1850
0.3237	93.0	1302	0.5836	0.3598	0.1854
0.3237	94.0	1316	0.5761	0.3588	0.1841
0.3237	95.0	1330	0.5822	0.3596	0.1848
0.3237	96.0	1344	0.5886	0.3592	0.1850
0.3237	97.0	1358	0.5696	0.3574	0.1830
0.3237	98.0	1372	0.5794	0.3588	0.1836
0.3237	99.0	1386	0.5768	0.3570	0.1837
0.2799	100.0	1400	0.5837	0.3578	0.1844
0.2799	101.0	1414	0.5697	0.3525	0.1826
0.2799	102.0	1428	0.5796	0.3566	0.1834
0.2799	103.0	1442	0.5712	0.3549	0.1825
0.2799	104.0	1456	0.5796	0.3555	0.1829
0.2799	105.0	1470	0.5759	0.3553	0.1835
0.2799	106.0	1484	0.5750	0.3562	0.1831
0.2799	107.0	1498	0.5650	0.3527	0.1823
0.2674	108.0	1512	0.5677	0.3499	0.1823
0.2674	109.0	1526	0.5699	0.3541	0.1826
0.2674	110.0	1540	0.5779	0.3555	0.1837
0.2674	111.0	1554	0.5792	0.3551	0.1834
0.2674	112.0	1568	0.5697	0.3574	0.1829
0.2674	113.0	1582	0.5852	0.3590	0.1839
0.2674	114.0	1596	0.5735	0.3537	0.1829
0.2611	115.0	1610	0.5774	0.3545	0.1832
0.2611	116.0	1624	0.5836	0.3555	0.1841
0.2611	117.0	1638	0.5750	0.3517	0.1832
0.2611	118.0	1652	0.5772	0.3521	0.1825
0.2611	119.0	1666	0.5793	0.3521	0.1831
0.2611	120.0	1680	0.5756	0.3517	0.1828
0.2611	121.0	1694	0.5794	0.3517	0.1830
0.2476	122.0	1708	0.5719	0.3521	0.1827
0.2476	123.0	1722	0.5804	0.3543	0.1830
0.2476	124.0	1736	0.5729	0.3539	0.1825
0.2476	125.0	1750	0.5874	0.3519	0.1832
0.2476	126.0	1764	0.5777	0.3533	0.1826
0.2476	127.0	1778	0.5762	0.3531	0.1822

Framework versions

• Transformers 4.28.0
• Pytorch 2.4.1+cu121
• Datasets 3.2.0
• Tokenizers 0.13.3