results

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

• Loss: 1.9132
• Accuracy: 0.6351
• F1: 0.6301
• Precision: 0.6315
• Recall: 0.6351

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: 2e-05
• 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
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.586	1.0	855	0.6272	0.6316	0.5289	0.5719	0.6316
0.6905	2.0	1710	0.5821	0.6520	0.5420	0.4804	0.6520
0.5642	3.0	2565	0.6012	0.6497	0.5396	0.4769	0.6497
0.6263	4.0	3420	0.6040	0.6526	0.5903	0.6403	0.6526
0.7556	5.0	4275	0.6088	0.6485	0.5406	0.4838	0.6485
0.4614	6.0	5130	0.6061	0.6614	0.6348	0.6523	0.6614
0.4232	7.0	5985	0.6184	0.6456	0.6401	0.6355	0.6456
0.519	8.0	6840	0.6186	0.6538	0.5449	0.4874	0.6538
0.4847	9.0	7695	0.6226	0.6421	0.5652	0.5948	0.6421
0.5505	10.0	8550	0.6857	0.6363	0.5903	0.6241	0.6363
0.5499	11.0	9405	0.6684	0.6491	0.6462	0.6470	0.6491
0.2869	12.0	10260	0.7500	0.6468	0.6341	0.6389	0.6468
0.2607	13.0	11115	0.9291	0.6351	0.6354	0.6357	0.6351
0.3495	14.0	11970	1.1298	0.6444	0.6415	0.6396	0.6444
0.2483	15.0	12825	1.3714	0.6246	0.6181	0.6320	0.6246
0.3764	16.0	13680	1.4325	0.6515	0.6454	0.6475	0.6515
0.373	17.0	14535	1.6233	0.6491	0.6473	0.6457	0.6491
0.0677	18.0	15390	1.7408	0.6450	0.6434	0.6429	0.6450
0.0665	19.0	16245	1.8868	0.6351	0.6296	0.6316	0.6351
0.0044	20.0	17100	1.9132	0.6351	0.6301	0.6315	0.6351

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.1+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1