w2v-bert-2.0-Fleurs_AMMI_AFRIVOICE_LRSC-ln-5hrs-v2

This model is a fine-tuned version of facebook/w2v-bert-2.0 on the None dataset. It achieves the following results on the evaluation set:

Model description

More information needed

More information needed

More information needed

The following hyperparameters were used during training:

learning_rate: 5e-05
train_batch_size: 8
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 100
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
1.6723	0.9949	98	0.5296	0.3535	0.1074
0.4866	2.0	197	0.5029	0.3311	0.0994
0.39	2.9949	295	0.4589	0.2949	0.0911
0.3091	4.0	394	0.4351	0.2753	0.0856
0.2712	4.9949	492	0.4382	0.2977	0.0907
0.2293	6.0	591	0.4217	0.2727	0.0838
0.1966	6.9949	689	0.4443	0.2709	0.0880
0.1638	8.0	788	0.4375	0.2514	0.0839
0.1413	8.9949	886	0.4777	0.2591	0.0876
0.1276	10.0	985	0.5044	0.2640	0.0812
0.1136	10.9949	1083	0.4985	0.2574	0.0806
0.0902	12.0	1182	0.5313	0.2340	0.0733
0.0963	12.9949	1280	0.5460	0.2497	0.0777
0.0731	14.0	1379	0.5303	0.2445	0.0756
0.0672	14.9949	1477	0.5373	0.2527	0.0788
0.0561	16.0	1576	0.5955	0.2481	0.0759
0.051	16.9949	1674	0.5695	0.2547	0.0809
0.044	18.0	1773	0.6118	0.2488	0.0758
0.045	18.9949	1871	0.5833	0.2624	0.0821
0.0368	20.0	1970	0.6394	0.2384	0.0743
0.0339	20.9949	2068	0.6359	0.2387	0.0751
0.0266	22.0	2167	0.6158	0.2337	0.0733
0.0277	22.9949	2265	0.5993	0.2526	0.0811
0.0234	24.0	2364	0.6488	0.2445	0.0762
0.0222	24.9949	2462	0.6444	0.2481	0.0761
0.0197	26.0	2561	0.6543	0.2480	0.0752
0.0144	26.9949	2659	0.6853	0.2431	0.0757