ramen-noodels/ast_delta_z

AST_finetined_buttonclick

This model is a fine-tuned version of MIT/ast-finetuned-audioset-10-10-0.4593 on a dataset collected un a UR3e robot. The model detects clicking sounds in the spectogram. It achieves the following results on the evaluation set:

• Loss: 0.0431
• Accuracy: 0.9886

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: 1e-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• 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
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.085	1.0	24	0.0804	0.9804
0.0206	2.0	48	0.0431	0.9886
0.009	3.0	72	0.0529	0.9853
0.0072	4.0	96	0.0615	0.9837
0.0018	5.0	120	0.0567	0.9865
0.0008	6.0	144	0.0646	0.9849
0.0005	7.0	168	0.0621	0.9869
0.0004	8.0	192	0.0637	0.9865
0.0002	9.0	216	0.0636	0.9865
0.0002	10.0	240	0.0635	0.9869

Framework versions

• Transformers 4.57.3
• Pytorch 2.7.1+cu118
• Datasets 4.4.1
• Tokenizers 0.22.1