wav2vec2-xls-r-Wolof-20-hours-kallaama-dataset

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2814
• Wer: 0.3377
• Cer: 0.1667

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: 0.0003
• 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
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
7.201	0.5548	200	3.5613	1.0	1.0
3.3219	1.1096	400	3.0794	1.0	1.0
3.1819	1.6644	600	3.0291	1.0	1.0
3.1057	2.2191	800	2.9537	1.0	1.0
2.7318	2.7739	1000	1.8572	0.9362	0.4899
2.0947	3.3287	1200	1.4787	0.7294	0.3568
1.8633	3.8835	1400	1.3119	0.6146	0.2935
1.7029	4.4383	1600	1.2306	0.6011	0.2828
1.6158	4.9931	1800	1.1482	0.5689	0.2747
1.5007	5.5479	2000	1.1997	0.5793	0.2849
1.4122	6.1026	2200	1.0793	0.5074	0.2387
1.3642	6.6574	2400	1.0829	0.5129	0.2411
1.3065	7.2122	2600	1.0104	0.4706	0.2218
1.2248	7.7670	2800	1.0017	0.4669	0.2178
1.2248	8.3218	3000	0.9914	0.4489	0.2121
1.1476	8.8766	3200	0.9413	0.4465	0.2140
1.1322	9.4313	3400	0.9384	0.4586	0.2195
1.0733	9.9861	3600	0.9237	0.4400	0.2089
1.0113	10.5409	3800	1.0012	0.4430	0.2150
1.0153	11.0957	4000	0.9331	0.4154	0.1976
0.9963	11.6505	4200	0.9500	0.4086	0.1984
0.9861	12.2053	4400	0.9105	0.4015	0.1923
0.9469	12.7601	4600	0.9050	0.3928	0.1879
0.936	13.3148	4800	0.9346	0.3951	0.1896
0.91	13.8696	5000	0.9387	0.4013	0.1928
0.9299	14.4244	5200	0.9697	0.4130	0.2004
0.882	14.9792	5400	0.9310	0.4155	0.1996
0.8354	15.5340	5600	1.0006	0.4064	0.1981
0.8471	16.0888	5800	0.9251	0.3829	0.1860
0.8236	16.6436	6000	0.9525	0.3801	0.1834
0.8194	17.1983	6200	0.9213	0.3772	0.1826
0.7891	17.7531	6400	0.9807	0.3718	0.1808
0.8135	18.3079	6600	0.9325	0.3809	0.1820
0.7832	18.8627	6800	0.8921	0.3859	0.1852
0.7663	19.4175	7000	0.9833	0.3939	0.1953
0.7663	19.9723	7200	0.9862	0.3834	0.1854
0.7227	20.5270	7400	1.0197	0.3758	0.1801
0.7249	21.0818	7600	0.9505	0.3671	0.1785
0.7113	21.6366	7800	0.9648	0.3672	0.1784
0.7116	22.1914	8000	0.9152	0.3709	0.1797
0.6927	22.7462	8200	0.9276	0.3636	0.1755
0.6947	23.3010	8400	0.9453	0.3734	0.1799
0.6665	23.8558	8600	0.9654	0.3677	0.1790
0.6684	24.4105	8800	1.0211	0.3761	0.1823
0.653	24.9653	9000	1.0406	0.3751	0.1796
0.6292	25.5201	9200	0.9922	0.3658	0.1792
0.6327	26.0749	9400	1.0176	0.3585	0.1756
0.6133	26.6297	9600	0.9768	0.3720	0.1801
0.6262	27.1845	9800	0.9729	0.3611	0.1756
0.596	27.7393	10000	0.9307	0.3579	0.1741
0.5914	28.2940	10200	1.0159	0.3582	0.1732
0.5767	28.8488	10400	1.0680	0.3684	0.1778
0.5615	29.4036	10600	1.0910	0.3587	0.1750
0.5531	29.9584	10800	1.0278	0.3639	0.1775
0.5467	30.5132	11000	1.0408	0.3580	0.1755
0.5372	31.0680	11200	1.0846	0.3524	0.1713
0.532	31.6227	11400	1.0610	0.3552	0.1740
0.5224	32.1775	11600	1.0826	0.3549	0.1734
0.5075	32.7323	11800	1.1042	0.3523	0.1711
0.5058	33.2871	12000	1.0994	0.3554	0.1724
0.493	33.8419	12200	1.0857	0.3516	0.1709
0.4661	34.3967	12400	1.1213	0.3507	0.1729
0.4764	34.9515	12600	1.1315	0.3491	0.1711
0.451	35.5062	12800	1.0818	0.3564	0.1745
0.4793	36.0610	13000	1.1029	0.3530	0.1731
0.4518	36.6158	13200	1.1308	0.3481	0.1709
0.4445	37.1706	13400	1.1243	0.3483	0.1718
0.4318	37.7254	13600	1.1597	0.3464	0.1705
0.4181	38.2802	13800	1.1771	0.3504	0.1709
0.4234	38.8350	14000	1.1590	0.3512	0.1714
0.402	39.3897	14200	1.2166	0.3430	0.1690
0.4118	39.9445	14400	1.2001	0.3460	0.1692
0.3908	40.4993	14600	1.1891	0.3423	0.1689
0.413	41.0541	14800	1.1905	0.3438	0.1686
0.3909	41.6089	15000	1.1846	0.3445	0.1694
0.3773	42.1637	15200	1.2198	0.3455	0.1699
0.3787	42.7184	15400	1.2136	0.3456	0.1686
0.3644	43.2732	15600	1.2274	0.3485	0.1700
0.3647	43.8280	15800	1.2453	0.3413	0.1678
0.3488	44.3828	16000	1.2309	0.3432	0.1684
0.3386	44.9376	16200	1.2498	0.3411	0.1678
0.3631	45.4924	16400	1.2552	0.3403	0.1670
0.34	46.0472	16600	1.2595	0.3418	0.1675
0.3347	46.6019	16800	1.2612	0.3419	0.1677
0.3396	47.1567	17000	1.2634	0.3394	0.1668
0.3396	47.7115	17200	1.2767	0.3384	0.1673
0.3369	48.2663	17400	1.2768	0.3389	0.1670
0.3353	48.8211	17600	1.2823	0.3383	0.1668
0.32	49.3759	17800	1.2837	0.3379	0.1668
0.3348	49.9307	18000	1.2814	0.3377	0.1667

Framework versions

• Transformers 4.44.1
• Pytorch 2.1.0+cu118
• Datasets 2.17.0
• Tokenizers 0.19.1