AST-finetuned-on-shEMO_speech

This model is a fine-tuned version of MIT/ast-finetuned-audioset-10-10-0.4593 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6988
• Accuracy: 0.7967

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• 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
0.8657	1.0	75	0.7066	0.7867
0.6951	2.0	150	0.6622	0.7567
0.3368	3.0	225	0.5851	0.8433
0.1414	4.0	300	0.7233	0.79
0.1011	5.0	375	0.8763	0.7967
0.0438	6.0	450	0.9009	0.8067
0.0108	7.0	525	1.0540	0.83
0.0033	8.0	600	1.0177	0.81
0.0003	9.0	675	1.1074	0.84
0.0113	10.0	750	1.1107	0.8433
0.0002	11.0	825	1.1273	0.8367
0.0001	12.0	900	1.1634	0.8333
0.0001	13.0	975	1.1502	0.84
0.0045	14.0	1050	1.1541	0.84
0.0039	15.0	1125	1.1550	0.84

Framework versions

• Transformers 4.34.1
• Pytorch 1.12.0+cu116
• Datasets 2.14.6
• Tokenizers 0.14.1