whale-call-detector

This model is a fine-tuned version of MIT/ast-finetuned-audioset-10-10-0.4593 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3433
• Accuracy: 0.925
• Precision: 0.9374
• Recall: 0.925
• F1: 0.9272
• F1 Water: 0.8889
• F1 Resident: 0.92
• F1 Transient: 0.9091
• F1 Humpback: 0.9524
• F1 Vessel: 1.0
• F1 Jingle: 0.8333
• F1 Human: 0.9412

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
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 0.1
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	F1 Water	F1 Resident	F1 Transient	F1 Humpback	F1 Vessel	F1 Jingle	F1 Human
0.4112	1.0	40	0.3943	0.8625	0.8847	0.8625	0.7935	0.8889	0.9362	0.6667	0.7778	0.9524	0.8333	0.9412
0.2625	2.0	80	0.5005	0.85	0.8648	0.85	0.8319	0.8421	0.9130	0.8	0.7826	0.8421	0.7273	0.9412
0.0173	3.0	120	0.3425	0.8875	0.9036	0.8875	0.8651	0.8571	0.92	0.8333	0.8421	0.8696	0.9231	0.9412
0.0015	4.0	160	0.3433	0.925	0.9374	0.925	0.9272	0.8889	0.92	0.9091	0.9524	1.0	0.8333	0.9412
0.0008	5.0	200	0.3720	0.9125	0.9238	0.9125	0.9272	0.8235	0.92	0.9091	0.9524	0.9524	0.8333	0.9412

Framework versions

• Transformers 5.8.1
• Pytorch 2.12.0+cu130
• Datasets 2.19.1
• Tokenizers 0.22.2