wav2vec2-base-finetuned-ks

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.9236
• Accuracy: 0.8129
• F1: 0.7991

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

BirdClef2023 (Top 20 species represented)

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 3e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.6415	1.0	1467	1.8215	0.4724	0.3482
1.4247	2.0	2934	1.3207	0.6457	0.5854
1.0871	3.0	4401	1.0206	0.7469	0.7164
0.4278	4.0	5868	0.9235	0.8006	0.7821
0.3165	5.0	7335	0.9236	0.8129	0.7991

Framework versions

• Transformers 4.31.0.dev0
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3