Lesson8-9results

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: 0.0043
• Accuracy: 0.9984
• F1-score: 0.9984
• Recall-score: 0.9984
• Precision-score: 0.9985

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: 16
• 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
2.2427	1.0	155	2.4337	0.4676	0.3990	0.4676	0.5162
1.4014	2.0	310	1.3308	0.8389	0.8209	0.8389	0.8401
0.5045	3.0	465	0.5602	0.9463	0.9346	0.9463	0.9295
0.1487	4.0	620	0.2187	0.9558	0.9447	0.9558	0.9398
0.0941	5.0	775	0.1194	0.9700	0.9656	0.9700	0.9811
2.4362	6.0	930	0.1300	0.9716	0.9717	0.9716	0.9809
0.0337	7.0	1085	0.0727	0.9874	0.9877	0.9874	0.9902
0.03	8.0	1240	0.0482	0.9921	0.9922	0.9921	0.9932
0.0075	9.0	1395	0.1213	0.9731	0.9730	0.9731	0.9766
0.0104	10.0	1550	0.0412	0.9905	0.9903	0.9905	0.9909
0.0065	11.0	1705	0.0383	0.9937	0.9937	0.9937	0.9941
0.0039	12.0	1860	0.0479	0.9874	0.9874	0.9874	0.9891
0.0037	13.0	2015	0.0930	0.9842	0.9838	0.9842	0.9868
0.0046	14.0	2170	0.0151	0.9968	0.9968	0.9968	0.9970
0.0054	15.0	2325	0.0085	0.9984	0.9984	0.9984	0.9985
0.0116	16.0	2480	0.0022	1.0	1.0	1.0	1.0
0.0028	17.0	2635	0.0022	1.0	1.0	1.0	1.0
0.0061	18.0	2790	0.0222	0.9953	0.9952	0.9953	0.9955
0.0024	19.0	2945	0.0048	0.9984	0.9984	0.9984	0.9985
0.0021	20.0	3100	0.0043	0.9984	0.9984	0.9984	0.9985

Framework versions

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