my_custom2_model

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: 0.4030
• Accuracy: 0.9167

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.1798	1.0	6	0.4439	0.75
0.2051	2.0	12	0.5505	0.5833
0.1612	3.0	18	0.1884	0.9167
0.2032	4.0	24	0.2759	0.9167
0.1803	5.0	30	0.5196	0.8333
0.0478	6.0	36	0.3214	0.9167
0.1159	7.0	42	0.3311	0.9167
0.031	8.0	48	0.6261	0.8333
0.0263	9.0	54	0.3536	0.9167
0.2505	10.0	60	0.3637	0.9167
0.018	11.0	66	0.3721	0.9167
0.0167	12.0	72	0.6487	0.8333
0.0154	13.0	78	0.7422	0.8333
0.0144	14.0	84	0.7221	0.8333
0.0129	15.0	90	0.5876	0.8333
0.0123	16.0	96	0.4041	0.9167
0.0118	17.0	102	0.4000	0.9167
0.0115	18.0	108	0.4015	0.9167
0.0112	19.0	114	0.4025	0.9167
0.011	20.0	120	0.4030	0.9167

Framework versions

• Transformers 4.50.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1