wav2vec2_three_classes_osd

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: 0.9518
• Accuracy: 0.8502

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.3576	1.0	3610	0.8598	0.7761
0.296	2.0	7220	0.6544	0.8245
0.2625	3.0	10830	0.6989	0.8249
0.2893	4.0	14441	0.7386	0.8329
0.2469	5.0	18051	0.8016	0.8433
0.2201	6.0	21661	0.7419	0.8444
0.2084	7.0	25271	0.8639	0.8449
0.2034	8.0	28882	0.8847	0.8499
0.1774	9.0	32492	0.9076	0.8443
0.1636	10.0	36102	0.9518	0.8502
0.1418	11.0	39712	1.0567	0.8469
0.1462	12.0	43323	1.0724	0.8449
0.1201	13.0	46933	1.0731	0.8448
0.1249	14.0	50543	1.1191	0.8440
0.1028	15.0	54153	1.0599	0.8486
0.1075	16.0	57764	1.1090	0.8465
0.1246	17.0	61374	1.1029	0.8497
0.088	18.0	64984	1.1654	0.8457
0.0875	19.0	68594	1.1944	0.8476
0.0695	20.0	72200	1.1755	0.8486

Framework versions

• Transformers 4.38.1
• Pytorch 2.2.0+cu121
• Datasets 2.17.1
• Tokenizers 0.15.2