search

SaoYear / SaoYear.github.io

0 stars 0 forks source link

EtherSpace of Nian #15

Open SaoYear opened 12 months ago

SaoYear commented 12 months ago

https://saoyear.github.io/post/finetune_ATST_SED/

Welcome to my site. Please feel free to contact me @ sao_year@126.com

martineghiazaryan commented 4 months ago

Hey, do you think the model could be suitable for mobile use with all the constraints?

SaoYear commented 4 months ago

@martineghiazaryan Hey, do you think the model could be suitable for mobile use with all the constraints?

Hi, currently I don't think so. The pretrained model is too large in both parameter amount and computational complexity.

The CRNN part could be adopted, although it is still much computationally heavier than the MobileNet, but it is possible. But it seems impossible to adopt ATST to the mobiles unless some quantilization techniques like VQ to be applied (Just guess, never tried before).

OrlandoWu11 commented 1 month ago

Hey, I am from China. I have a great interest of this job! Unfortunately, I can not download all the dataset of dcase, could u please share it all through Chinese cloud disk? Thanks

SaoYear commented 1 month ago

@OrlandoWu11 Hey, I am from China. I have a great interest of this job! Unfortunately, I can not download all the dataset of dcase, could u please share it all through Chinese cloud disk? Thanks

Hi Wu,

The full dataset has been shared in a Chinese cloud disk. You could find it in either 知乎 or in the Github page of ATST-SED.