Lesson1results

This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0149
• Accuracy: 0.9962
• F1-score: 0.9962
• Recall-score: 0.9962
• Precision-score: 0.9962

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: 5e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1-score	Recall-score	Precision-score
1.1736	1.0	278	0.9850	0.7898	0.7552	0.7898	0.7813
0.4185	2.0	556	0.4326	0.9106	0.8991	0.9106	0.9062
0.3854	3.0	834	0.2507	0.9363	0.9335	0.9363	0.9409
0.2509	4.0	1112	0.1460	0.9666	0.9665	0.9666	0.9673
0.107	5.0	1390	0.1278	0.9641	0.9640	0.9641	0.9689
0.3585	6.0	1668	0.1188	0.9758	0.9758	0.9758	0.9764
0.2611	7.0	1946	0.1148	0.9704	0.9702	0.9704	0.9722
0.2493	8.0	2224	0.0638	0.9824	0.9824	0.9824	0.9828
0.0351	9.0	2502	0.0492	0.9887	0.9887	0.9887	0.9890
0.4708	10.0	2780	0.0479	0.9883	0.9883	0.9883	0.9885
0.2958	11.0	3058	0.0561	0.9865	0.9865	0.9865	0.9870
0.138	12.0	3336	0.0308	0.9916	0.9916	0.9916	0.9918
0.0525	13.0	3614	0.0226	0.9944	0.9944	0.9944	0.9944
0.0332	14.0	3892	0.0293	0.9916	0.9916	0.9916	0.9920
0.0332	15.0	4170	0.0202	0.9953	0.9953	0.9953	0.9953
0.339	16.0	4448	0.0210	0.9955	0.9955	0.9955	0.9955
0.211	17.0	4726	0.0218	0.9959	0.9959	0.9959	0.9960
0.0017	18.0	5004	0.0181	0.9964	0.9964	0.9964	0.9964
0.1646	19.0	5282	0.0166	0.9959	0.9959	0.9959	0.9960
0.0014	20.0	5560	0.0149	0.9962	0.9962	0.9962	0.9962

Framework versions

• Transformers 4.42.3
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1