testlaibasettsgopdata

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: 0.0930
• Wer: 0.1682

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
6.2716	1.05	500	3.0550	1.0
1.8262	2.11	1000	0.2669	0.3023
0.5469	3.16	1500	0.1809	0.2281
0.3541	4.21	2000	0.1541	0.2185
0.3367	5.26	2500	0.1432	0.2054
0.2792	6.32	3000	0.1218	0.2023
0.2411	7.37	3500	0.1136	0.2029
0.2041	8.42	4000	0.1423	0.2025
0.2262	9.47	4500	0.1294	0.1968
0.1921	10.53	5000	0.1237	0.1952
0.1877	11.58	5500	0.1043	0.1890
0.176	12.63	6000	0.1272	0.1935
0.1236	13.68	6500	0.1352	0.1902
0.1473	14.74	7000	0.1257	0.1874
0.1748	15.79	7500	0.1190	0.1854
0.1147	16.84	8000	0.1213	0.1914
0.1508	17.89	8500	0.1262	0.1813
0.1061	18.95	9000	0.1148	0.1802
0.1182	20.0	9500	0.1034	0.1758
0.1144	21.05	10000	0.1123	0.1769
0.0885	22.11	10500	0.1043	0.1735
0.0797	23.16	11000	0.1004	0.1712
0.0729	24.21	11500	0.1045	0.1703
0.0718	25.26	12000	0.1064	0.1712
0.0668	26.32	12500	0.1050	0.1687
0.0599	27.37	13000	0.0965	0.1677
0.0702	28.42	13500	0.0930	0.1682
0.0942	29.47	14000	0.0959	0.1674

Framework versions

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