wav2vec2-base-finetuned-mednames1

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: 0.5263
• Accuracy: 0.9727

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: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2881	1.0	22	2.2690	0.2273
2.2234	2.0	44	2.1631	0.2273
1.9756	3.0	66	1.9277	0.5182
1.7191	4.0	88	1.6293	0.6818
1.4663	5.0	110	1.3919	0.7545
1.27	6.0	132	1.1689	0.8182
1.1112	7.0	154	1.0144	0.8364
0.9623	8.0	176	0.9137	0.8545
0.8764	9.0	198	0.7901	0.8909
0.7776	10.0	220	0.7229	0.8727
0.7266	11.0	242	0.6335	0.9
0.6379	12.0	264	0.5848	0.9636
0.6121	13.0	286	0.5509	0.9273
0.5732	14.0	308	0.5263	0.9727
0.5579	15.0	330	0.5221	0.9727

Framework versions

• Transformers 4.27.2
• Pytorch 1.11.0+cu102
• Datasets 2.10.1
• Tokenizers 0.13.2