wav2vec2-large-xlsr-coraa-exp-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: 0.5470
• Wer: 0.3417
• Cer: 0.1785

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
37.9224	1.0	14	24.6578	1.0	0.9618
37.9224	2.0	28	7.1453	1.0	0.9619
37.9224	3.0	42	4.4391	1.0	0.9619
37.9224	4.0	56	3.9092	1.0	0.9619
37.9224	5.0	70	3.6835	1.0	0.9619
37.9224	6.0	84	3.5223	1.0	0.9619
37.9224	7.0	98	3.3716	1.0	0.9619
9.0651	8.0	112	3.2723	1.0	0.9619
9.0651	9.0	126	3.1860	1.0	0.9619
9.0651	10.0	140	3.1461	1.0	0.9619
9.0651	11.0	154	3.1368	1.0	0.9619
9.0651	12.0	168	3.0961	1.0	0.9619
9.0651	13.0	182	3.0767	1.0	0.9619
9.0651	14.0	196	3.0509	1.0	0.9619
3.0601	15.0	210	3.0871	1.0	0.9619
3.0601	16.0	224	3.0415	1.0	0.9619
3.0601	17.0	238	3.0330	1.0	0.9619
3.0601	18.0	252	3.0192	1.0	0.9619
3.0601	19.0	266	3.0266	1.0	0.9619
3.0601	20.0	280	3.0243	1.0	0.9619
3.0601	21.0	294	3.0106	1.0	0.9619
2.9552	22.0	308	3.0053	1.0	0.9619
2.9552	23.0	322	2.9986	1.0	0.9619
2.9552	24.0	336	3.0030	1.0	0.9619
2.9552	25.0	350	2.9950	1.0	0.9619
2.9552	26.0	364	3.0058	1.0	0.9619
2.9552	27.0	378	2.9943	1.0	0.9619
2.9552	28.0	392	2.9845	1.0	0.9619
2.9213	29.0	406	2.9713	1.0	0.9619
2.9213	30.0	420	2.9485	1.0	0.9619
2.9213	31.0	434	2.9415	1.0	0.9619
2.9213	32.0	448	2.8913	1.0	0.9619
2.9213	33.0	462	2.8057	1.0	0.9612
2.9213	34.0	476	2.6984	1.0	0.9599
2.9213	35.0	490	2.5785	1.0	0.9067
2.7804	36.0	504	2.3545	1.0	0.7929
2.7804	37.0	518	2.0433	1.0	0.5933
2.7804	38.0	532	1.7438	1.0	0.4701
2.7804	39.0	546	1.4659	1.0	0.4139
2.7804	40.0	560	1.2873	0.9929	0.3840
2.7804	41.0	574	1.1588	0.9315	0.3387
2.7804	42.0	588	1.0163	0.7395	0.2706
1.6517	43.0	602	0.9399	0.5331	0.2258
1.6517	44.0	616	0.9131	0.4929	0.2167
1.6517	45.0	630	0.8352	0.4770	0.2114
1.6517	46.0	644	0.8115	0.4555	0.2084
1.6517	47.0	658	0.7850	0.4403	0.2038
1.6517	48.0	672	0.7574	0.4356	0.2019
1.6517	49.0	686	0.7238	0.4248	0.1989
0.7966	50.0	700	0.7132	0.4130	0.1960
0.7966	51.0	714	0.7054	0.4128	0.1963
0.7966	52.0	728	0.7119	0.4134	0.1990
0.7966	53.0	742	0.6793	0.3990	0.1945
0.7966	54.0	756	0.6718	0.3944	0.1932
0.7966	55.0	770	0.6718	0.4013	0.1949
0.7966	56.0	784	0.6831	0.3976	0.1965
0.7966	57.0	798	0.6400	0.3870	0.1916
0.5799	58.0	812	0.6423	0.3844	0.1906
0.5799	59.0	826	0.6394	0.3834	0.1908
0.5799	60.0	840	0.6574	0.3785	0.1924
0.5799	61.0	854	0.6321	0.3816	0.1918
0.5799	62.0	868	0.6306	0.3801	0.1913
0.5799	63.0	882	0.6433	0.3799	0.1916
0.5799	64.0	896	0.6342	0.3811	0.1896
0.445	65.0	910	0.6212	0.3811	0.1904
0.445	66.0	924	0.6164	0.3789	0.1895
0.445	67.0	938	0.6006	0.3732	0.1872
0.445	68.0	952	0.6054	0.3746	0.1891
0.445	69.0	966	0.6245	0.3722	0.1894
0.445	70.0	980	0.6090	0.3688	0.1878
0.445	71.0	994	0.6073	0.3669	0.1876
0.3746	72.0	1008	0.5989	0.3708	0.1889
0.3746	73.0	1022	0.5968	0.3681	0.1874
0.3746	74.0	1036	0.5946	0.3659	0.1870
0.3746	75.0	1050	0.5874	0.3623	0.1864
0.3746	76.0	1064	0.5928	0.3639	0.1870
0.3746	77.0	1078	0.5889	0.3681	0.1882
0.3746	78.0	1092	0.5723	0.3683	0.1864
0.3543	79.0	1106	0.5928	0.3657	0.1863
0.3543	80.0	1120	0.5832	0.3649	0.1855
0.3543	81.0	1134	0.5785	0.3645	0.1849
0.3543	82.0	1148	0.5877	0.3580	0.1842
0.3543	83.0	1162	0.5870	0.3627	0.1853
0.3543	84.0	1176	0.5738	0.3618	0.1846
0.3543	85.0	1190	0.5641	0.3576	0.1816
0.3207	86.0	1204	0.5728	0.3566	0.1821
0.3207	87.0	1218	0.5706	0.3560	0.1817
0.3207	88.0	1232	0.5607	0.3570	0.1813
0.3207	89.0	1246	0.5644	0.3557	0.1817
0.3207	90.0	1260	0.5660	0.3582	0.1824
0.3207	91.0	1274	0.5688	0.3566	0.1829
0.3207	92.0	1288	0.5635	0.3541	0.1807
0.2984	93.0	1302	0.5663	0.3503	0.1814
0.2984	94.0	1316	0.5515	0.3543	0.1807
0.2984	95.0	1330	0.5563	0.3517	0.1803
0.2984	96.0	1344	0.5618	0.3509	0.1809
0.2984	97.0	1358	0.5554	0.3517	0.1807
0.2984	98.0	1372	0.5606	0.3529	0.1809
0.2984	99.0	1386	0.5597	0.3511	0.1813
0.2622	100.0	1400	0.5628	0.3505	0.1812
0.2622	101.0	1414	0.5564	0.3495	0.1799
0.2622	102.0	1428	0.5626	0.3484	0.1811
0.2622	103.0	1442	0.5556	0.3470	0.1800
0.2622	104.0	1456	0.5603	0.3464	0.1799
0.2622	105.0	1470	0.5571	0.3454	0.1799
0.2622	106.0	1484	0.5618	0.3466	0.1799
0.2622	107.0	1498	0.5519	0.3440	0.1787
0.2519	108.0	1512	0.5541	0.3440	0.1790
0.2519	109.0	1526	0.5574	0.3464	0.1794
0.2519	110.0	1540	0.5590	0.3454	0.1801
0.2519	111.0	1554	0.5530	0.3448	0.1796
0.2519	112.0	1568	0.5501	0.3438	0.1792
0.2519	113.0	1582	0.5595	0.3448	0.1799
0.2519	114.0	1596	0.5536	0.3446	0.1801
0.245	115.0	1610	0.5480	0.3432	0.1789
0.245	116.0	1624	0.5623	0.3486	0.1798
0.245	117.0	1638	0.5496	0.3427	0.1790
0.245	118.0	1652	0.5552	0.3421	0.1789
0.245	119.0	1666	0.5558	0.3438	0.1787
0.245	120.0	1680	0.5524	0.3425	0.1783
0.245	121.0	1694	0.5582	0.3421	0.1786
0.2322	122.0	1708	0.5534	0.3425	0.1786
0.2322	123.0	1722	0.5596	0.3464	0.1801
0.2322	124.0	1736	0.5486	0.3432	0.1790
0.2322	125.0	1750	0.5581	0.3425	0.1792
0.2322	126.0	1764	0.5470	0.3417	0.1785
0.2322	127.0	1778	0.5544	0.3413	0.1781
0.2322	128.0	1792	0.5501	0.3436	0.1781
0.2324	129.0	1806	0.5518	0.3440	0.1782
0.2324	130.0	1820	0.5511	0.3389	0.1775
0.2324	131.0	1834	0.5584	0.3417	0.1782
0.2324	132.0	1848	0.5493	0.3373	0.1775
0.2324	133.0	1862	0.5506	0.3395	0.1777
0.2324	134.0	1876	0.5543	0.3409	0.1782
0.2324	135.0	1890	0.5589	0.3399	0.1781
0.2077	136.0	1904	0.5556	0.3391	0.1778
0.2077	137.0	1918	0.5555	0.3407	0.1778
0.2077	138.0	1932	0.5501	0.3391	0.1774
0.2077	139.0	1946	0.5544	0.3375	0.1772
0.2077	140.0	1960	0.5554	0.3387	0.1773
0.2077	141.0	1974	0.5504	0.3381	0.1772
0.2077	142.0	1988	0.5484	0.3383	0.1770
0.2089	143.0	2002	0.5519	0.3385	0.1772
0.2089	144.0	2016	0.5532	0.3391	0.1772
0.2089	145.0	2030	0.5530	0.3397	0.1775
0.2089	146.0	2044	0.5551	0.3397	0.1775

Framework versions

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