search

Deci-AI / super-gradients

Easily train or fine-tune SOTA computer vision models with one open source training library. The home of Yolo-NAS.
https://www.supergradients.com
Apache License 2.0
4.59k stars 510 forks source link

Any paper or document about how to reproduce the results of the pretrained weights? #1590

Closed stereomatchingkiss closed 1 year ago

stereomatchingkiss commented 1 year ago

💡 Your Question

I would like to retrain the yolo nas S model on the coco dataset, do this project has any document or paper to show us how to replicate the results? Thanks

Versions

3.3.1

BloodAxe commented 1 year ago

We provide a coco2017_yolo_nas_s.yaml recipe to train a Yolo-NAS pose from scratch on COCO dataset. This is NOT recipe to reproduce our results, but a good starting point if you want to train your own version of YoloNAS that is not bound to original YoloNAS weights license.

We trained YoloNAS in a few staged, including pre-training on O365 and further KD on COCO, which is too complicated and at the moment there are no plans to release it.