wav2vec2_classifier_arabic

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: 0.6082
• Accuracy: 0.8571
• Precision: 0.8713
• Recall: 0.8571
• F1: 0.8549
• Binary: 0.9002

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: 6.255516869030134e-05
• train_batch_size: 64
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• training_steps: 2000

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Binary
4.4473	1.0	109	4.4217	0.0257	0.0267	0.0257	0.0125	0.1703
4.2304	2.0	219	4.0329	0.1027	0.0299	0.1027	0.0402	0.3658
3.9161	3.0	328	3.6520	0.1637	0.1035	0.1637	0.0937	0.4124
3.5075	4.0	438	3.2131	0.3098	0.2155	0.3098	0.2227	0.5165
3.1463	5.0	547	2.7609	0.4494	0.3587	0.4494	0.3650	0.6128
2.7101	6.0	657	2.2656	0.6148	0.5661	0.6148	0.5558	0.7300
2.3391	7.0	766	1.8535	0.6918	0.6991	0.6918	0.6567	0.7841
1.977	8.0	876	1.4528	0.7640	0.7554	0.7640	0.7389	0.8350
1.7077	9.0	985	1.2586	0.7673	0.7744	0.7673	0.7504	0.8353
1.4602	10.0	1095	1.0169	0.8154	0.8373	0.8154	0.8054	0.8711
1.2622	11.0	1204	0.8802	0.8250	0.8503	0.8250	0.8227	0.8772
1.1108	12.0	1314	0.7670	0.8315	0.8572	0.8315	0.8259	0.8812
0.9983	13.0	1423	0.7210	0.8427	0.8666	0.8427	0.8412	0.8905
0.8929	14.0	1533	0.6566	0.8411	0.8594	0.8411	0.8375	0.8894
0.8259	15.0	1642	0.6354	0.8539	0.8684	0.8539	0.8509	0.8979
0.7617	16.0	1752	0.6634	0.8507	0.8669	0.8507	0.8483	0.8957
0.7255	17.0	1861	0.6046	0.8571	0.8689	0.8571	0.8547	0.9006
0.6853	18.0	1971	0.6096	0.8587	0.8757	0.8587	0.8555	0.9013
0.666	18.26	2000	0.6082	0.8571	0.8713	0.8571	0.8549	0.9002

Framework versions

• Transformers 4.33.3
• Pytorch 2.2.1
• Datasets 2.18.0
• Tokenizers 0.13.3