audio-commands

This model is a fine-tuned version of MIT/ast-finetuned-speech-commands-v2 on the speech_commands dataset. It achieves the following results on the evaluation set:

• Loss: 0.3977
• Accuracy: 0.9256

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
0.0581	1.0	663	0.4816	0.8975
0.0454	2.0	1326	0.4184	0.9024
0.0404	3.0	1989	0.4361	0.9010
0.025	4.0	2653	0.4368	0.9016
0.0169	5.0	3316	0.3692	0.9173
0.0173	6.0	3979	0.4131	0.9173
0.0096	7.0	4642	0.3800	0.9177
0.0022	8.0	5306	0.3535	0.9264
0.0031	9.0	5969	0.3241	0.9315
0.0008	10.0	6632	0.3697	0.9236
0.0002	11.0	7295	0.4189	0.9173
0.001	12.0	7959	0.3206	0.9287
0.0003	13.0	8622	0.3794	0.9205
0.0003	14.0	9285	0.3999	0.9199
0.0	15.0	9948	0.4002	0.9220
0.0	16.0	10612	0.3896	0.9248
0.0001	17.0	11275	0.3930	0.9248
0.0	18.0	11938	0.3952	0.9254
0.0	19.0	12601	0.3971	0.9254
0.0	19.99	13260	0.3977	0.9256

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.2+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2