Tune for COCO - Githubissues

eriklindernoren / PyTorch-YOLOv3

Minimal PyTorch implementation of YOLOv3

GNU General Public License v3.0

7.29k stars 2.64k forks source link

Tune for COCO #826

Open Flova opened 1 year ago

Flova commented 1 year ago

Proposed changes

tune hyperparams
scale based on mini batch size
linear instead of step wise decay
use sgd
only log once per optimizer step

Related issues

818
820

Necessary checks

[ ] Update poetry package version semantically
[x] Write documentation
[x] Create issues for future work
[x] Test on your machine

J-LINC commented 1 year ago

hi ，Regarding what you mentioned:linear instead of step wise decay, I have reviewed your code and found that there is only one learning rate decay strategy, which is to adjust according to steps. How can I perform linear decay or other decay methods as for using sgd ,i think you are right for most trains and i successed when i using sgd in trianing bdd100k