gopdatastt_add_transformer

This model is a fine-tuned version of facebook/wav2vec2-base-960h on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0920
• Wer: 0.1617

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.1709	1.05	500	0.1453	0.2194
0.3131	2.11	1000	0.1094	0.2055
0.276	3.16	1500	0.1198	0.1998
0.2416	4.21	2000	0.1873	0.2026
0.2093	5.26	2500	0.1392	0.1974
0.1987	6.32	3000	0.1123	0.1944
0.1714	7.37	3500	0.1089	0.1890
0.1634	8.42	4000	0.1007	0.1863
0.1459	9.47	4500	0.1340	0.1864
0.1461	10.53	5000	0.1016	0.1874
0.1316	11.58	5500	0.1110	0.1891
0.1318	12.63	6000	0.0942	0.1855
0.1084	13.68	6500	0.0992	0.1827
0.1064	14.74	7000	0.1010	0.1801
0.1059	15.79	7500	0.1173	0.1834
0.094	16.84	8000	0.1096	0.1815
0.0918	17.89	8500	0.1046	0.1780
0.0874	18.95	9000	0.1103	0.1788
0.0813	20.0	9500	0.1065	0.1768
0.0753	21.05	10000	0.0997	0.1747
0.0729	22.11	10500	0.1053	0.1748
0.0655	23.16	11000	0.1042	0.1726
0.0647	24.21	11500	0.0960	0.1746
0.0581	25.26	12000	0.1060	0.1733
0.0573	26.32	12500	0.0972	0.1706
0.0524	27.37	13000	0.0963	0.1725
0.0577	28.42	13500	0.0920	0.1696
0.0488	29.47	14000	0.0942	0.1686

Framework versions

• Transformers 4.17.0
• Pytorch 2.5.1+cu121
• Datasets 1.18.3
• Tokenizers 0.20.3