w2v-bert-2.0-Fleurs_AMMI_AFRIVOICE_LRSC-ln-5hrs-v1

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
1.8874	0.9949	98	0.6403	0.5429	0.1657
0.4899	2.0	197	0.4921	0.3300	0.1001
0.3892	2.9949	295	0.4608	0.3314	0.1019
0.3259	4.0	394	0.4729	0.3080	0.0942
0.2863	4.9949	492	0.4495	0.3156	0.0951
0.2333	6.0	591	0.4269	0.2624	0.0808
0.2059	6.9949	689	0.4365	0.2609	0.0839
0.1722	8.0	788	0.4346	0.2552	0.0825
0.1551	8.9949	886	0.4134	0.2468	0.0766
0.1318	10.0	985	0.4794	0.2631	0.0811
0.1189	10.9949	1083	0.5191	0.2530	0.0796
0.1004	12.0	1182	0.5311	0.2689	0.0794
0.0959	12.9949	1280	0.5502	0.2535	0.0778
0.0831	14.0	1379	0.5060	0.2476	0.0757
0.0679	14.9949	1477	0.5023	0.2517	0.0830
0.0617	16.0	1576	0.5279	0.2403	0.0757
0.0562	16.9949	1674	0.6012	0.2411	0.0761
0.0496	18.0	1773	0.6263	0.2423	0.0755
0.0442	18.9949	1871	0.5991	0.2581	0.0794
0.0401	20.0	1970	0.6323	0.2412	0.0762
0.0329	20.9949	2068	0.6417	0.2326	0.0735
0.0266	22.0	2167	0.6279	0.2381	0.0756
0.0255	22.9949	2265	0.5834	0.2470	0.0772
0.0214	24.0	2364	0.6781	0.2364	0.0735
0.0217	24.9949	2462	0.6253	0.2398	0.0752
0.0163	26.0	2561	0.6940	0.2427	0.0813
0.0363	26.9949	2659	0.6632	0.2363	0.0756
0.0182	28.0	2758	0.6094	0.2363	0.0766
0.014	28.9949	2856	0.6928	0.2438	0.0770
0.0157	30.0	2955	0.6863	0.2422	0.0768
0.0121	30.9949	3053	0.6643	0.2469	0.0788