wav2vec2-base-finetuned-mednames

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

• Loss: 0.1443
• Accuracy: 1.0

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: 5
• eval_batch_size: 5
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 20
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2789	1.0	36	2.2639	0.2222
2.1925	2.0	72	2.1404	0.3833
1.8763	3.0	108	1.7607	0.5556
1.5035	4.0	144	1.3599	0.7667
1.0598	5.0	180	0.9607	0.8778
0.8303	6.0	216	0.7327	0.8722
0.6418	7.0	252	0.5544	0.8778
0.4481	8.0	288	0.4377	0.9722
0.3438	9.0	324	0.2963	0.9833
0.1831	10.0	360	0.1443	1.0
0.1129	11.0	396	0.0890	1.0
0.0877	12.0	432	0.0664	1.0
0.0657	13.0	468	0.0509	1.0
0.0528	14.0	504	0.0406	1.0
0.0428	15.0	540	0.0388	1.0
0.0369	16.0	576	0.0526	0.9944
0.0858	17.0	612	0.0539	0.9944
0.0294	18.0	648	0.0231	1.0
0.0272	19.0	684	0.0204	1.0
0.0247	20.0	720	0.0188	1.0
0.0227	21.0	756	0.0172	1.0
0.0211	22.0	792	0.0161	1.0
0.0195	23.0	828	0.0152	1.0
0.0684	24.0	864	0.0144	1.0
0.0175	25.0	900	0.0139	1.0
0.0756	26.0	936	0.0134	1.0
0.017	27.0	972	0.0131	1.0
0.0163	28.0	1008	0.0128	1.0
0.0158	29.0	1044	0.0126	1.0
0.0159	30.0	1080	0.0126	1.0

Framework versions

• Transformers 4.28.1
• Pytorch 1.11.0+cu102
• Datasets 2.11.0
• Tokenizers 0.13.3