wav2vec2-xls-r-wolof-mixed-75-hours

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

• Loss: 1.0600
• Wer: 0.4136
• Cer: 0.1413

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: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 36

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
15.2262	0.3478	400	3.9480	1.0	1.0
6.2701	0.6957	800	3.1176	1.0	1.0
4.3901	1.0435	1200	1.1108	0.7271	0.2440
2.3527	1.3913	1600	0.8827	0.5767	0.1992
2.0303	1.7391	2000	0.7998	0.5375	0.1872
1.8695	2.0870	2400	0.7999	0.5131	0.1835
1.7196	2.4348	2800	0.7045	0.5181	0.1821
1.7028	2.7826	3200	0.7431	0.4909	0.1770
1.6103	3.1304	3600	0.7454	0.4996	0.1794
1.5956	3.4783	4000	0.7233	0.5003	0.1761
1.5677	3.8261	4400	0.6952	0.5044	0.1782
1.4969	4.1739	4800	0.7432	0.5196	0.1815
1.4505	4.5217	5200	0.6875	0.4674	0.1657
1.4267	4.8696	5600	0.6290	0.4924	0.1717
1.3537	5.2174	6000	0.6311	0.4858	0.1718
1.3208	5.5652	6400	0.6676	0.4720	0.1659
1.3139	5.9130	6800	0.6329	0.5023	0.1729
1.2517	6.2609	7200	0.6445	0.4987	0.1753
1.2352	6.6087	7600	0.6670	0.4629	0.1645
1.2174	6.9565	8000	0.6350	0.4776	0.1645
1.1423	7.3043	8400	0.6674	0.4537	0.1584
1.1593	7.6522	8800	0.5977	0.4508	0.1586
1.1651	8.0	9200	0.6381	0.4581	0.1598
1.0786	8.3478	9600	0.6538	0.4459	0.1573
1.0895	8.6957	10000	0.6510	0.4759	0.1635
1.0842	9.0435	10400	0.6333	0.4545	0.1587
0.9992	9.3913	10800	0.6645	0.4452	0.1547
1.0316	9.7391	11200	0.5950	0.4535	0.1582
1.0176	10.0870	11600	0.6691	0.4538	0.1591
0.9826	10.4348	12000	0.6146	0.4576	0.1612
0.9581	10.7826	12400	0.6362	0.4366	0.1518
0.9315	11.1304	12800	0.6494	0.4671	0.1598
0.9231	11.4783	13200	0.6859	0.4285	0.1516
0.9268	11.8261	13600	0.6601	0.4287	0.1528
0.9014	12.1739	14000	0.6848	0.4398	0.1536
0.8824	12.5217	14400	0.6399	0.4296	0.1505
0.8893	12.8696	14800	0.6754	0.4327	0.1555
0.8332	13.2174	15200	0.6464	0.4399	0.1555
0.8197	13.5652	15600	0.6878	0.4390	0.1533
0.8641	13.9130	16000	0.6685	0.4387	0.1536
0.8037	14.2609	16400	0.6413	0.4451	0.1521
0.8075	14.6087	16800	0.6770	0.4314	0.1501
0.7989	14.9565	17200	0.6403	0.4527	0.1578
0.777	15.3043	17600	0.6591	0.4323	0.1515
0.7647	15.6522	18000	0.6981	0.4357	0.1540
0.7668	16.0	18400	0.6699	0.4248	0.1477
0.7258	16.3478	18800	0.6743	0.4424	0.1512
0.735	16.6957	19200	0.6855	0.4392	0.1519
0.7163	17.0435	19600	0.7157	0.4334	0.1515
0.7059	17.3913	20000	0.7152	0.4391	0.1519
0.6953	17.7391	20400	0.6298	0.4291	0.1504
0.6923	18.0870	20800	0.7449	0.4373	0.1517
0.6654	18.4348	21200	0.6600	0.4243	0.1487
0.6776	18.7826	21600	0.7403	0.4291	0.1484
0.659	19.1304	22000	0.7134	0.4368	0.1509
0.6427	19.4783	22400	0.7312	0.4296	0.1491
0.6386	19.8261	22800	0.7457	0.4403	0.1538
0.6375	20.1739	23200	0.7278	0.4417	0.1524
0.6202	20.5217	23600	0.7500	0.4336	0.1493
0.6212	20.8696	24000	0.7423	0.4229	0.1480
0.6119	21.2174	24400	0.7775	0.4273	0.1485
0.588	21.5652	24800	0.7921	0.4313	0.1484
0.6015	21.9130	25200	0.7427	0.4240	0.1460
0.5641	22.2609	25600	0.8377	0.4322	0.1491
0.5718	22.6087	26000	0.7952	0.4293	0.1499
0.5936	22.9565	26400	0.8305	0.4365	0.1496
0.5587	23.3043	26800	0.8194	0.4386	0.1513
0.5488	23.6522	27200	0.7958	0.4288	0.1470
0.5532	24.0	27600	0.8417	0.4227	0.1451
0.5325	24.3478	28000	0.8885	0.4282	0.1475
0.5233	24.6957	28400	0.8220	0.4334	0.1489
0.5247	25.0435	28800	0.9235	0.4281	0.1466
0.5185	25.3913	29200	0.8981	0.4330	0.1493
0.5099	25.7391	29600	0.8012	0.4311	0.1479
0.5066	26.0870	30000	0.9317	0.4199	0.1446
0.489	26.4348	30400	0.9119	0.4260	0.1454
0.4989	26.7826	30800	0.8681	0.4209	0.1447
0.4855	27.1304	31200	0.8741	0.4313	0.1467
0.4724	27.4783	31600	0.8952	0.4205	0.1426
0.4715	27.8261	32000	0.8722	0.4275	0.1457
0.4654	28.1739	32400	0.9719	0.4321	0.1472
0.4678	28.5217	32800	0.9647	0.4250	0.1453
0.4599	28.8696	33200	0.9411	0.4300	0.1462
0.4374	29.2174	33600	0.9919	0.4231	0.1447
0.4486	29.5652	34000	0.9928	0.4220	0.1432
0.438	29.9130	34400	0.9597	0.4276	0.1453
0.435	30.2609	34800	0.9759	0.4273	0.1455
0.4252	30.6087	35200	0.9845	0.4267	0.1459
0.4173	30.9565	35600	1.0023	0.4255	0.1447
0.4281	31.3043	36000	1.0498	0.4223	0.1443
0.4079	31.6522	36400	1.0335	0.4206	0.1434
0.4176	32.0	36800	0.9958	0.4217	0.1435
0.4013	32.3478	37200	1.0534	0.4209	0.1439
0.401	32.6957	37600	1.0248	0.4145	0.1416
0.4081	33.0435	38000	1.0508	0.4165	0.1418
0.3951	33.3913	38400	1.0494	0.4182	0.1422
0.3852	33.7391	38800	1.0373	0.4138	0.1411
0.3985	34.0870	39200	1.0426	0.4168	0.1424
0.3823	34.4348	39600	1.0586	0.4169	0.1423
0.3859	34.7826	40000	1.0514	0.4179	0.1422
0.3948	35.1304	40400	1.0770	0.4145	0.1414
0.3637	35.4783	40800	1.0568	0.4145	0.1411
0.371	35.8261	41200	1.0600	0.4136	0.1413

Framework versions

• Transformers 4.46.1
• Pytorch 2.5.1+cu124
• Datasets 2.17.0
• Tokenizers 0.20.3