search

fschmid56 / EfficientAT

This repository aims at providing efficient CNNs for Audio Tagging. We provide AudioSet pre-trained models ready for downstream training and extraction of audio embeddings.
MIT License
218 stars 41 forks source link

ImageNet training results #33

Closed hbellafkir closed 2 weeks ago

hbellafkir commented 2 weeks ago

Thank you for sharing your work. Could you also provide the evaluation results from the ImageNet training?

fschmid56 commented 2 weeks ago

Hi! ImageNet pre-training was mostly done by a colleague of mine. However, I remember that we used this repo.

And I also found some old logs with some results:

MN10: 'top_1': 0.7186, 'top_5': 0.9076 DyMN10: 'top_1': 0.7320, 'top_5': 0.9119

However, these results might not exactly match the checkpoints uploaded to this repo. But the results should be somewhat similar if you need them for a sanity check.

hbellafkir commented 2 weeks ago

Thanks for the quick reply.