Jay Desai comments

Results 45 comments of


                                            Jay Desai

Is it possible training SSD or YOLOv3 in fp16?

+1 @Jerryzcn you can double the batch size 2x fp32 and increase the learning rate as well when you use amp making training much faster!

A reimplementation of MixNet in Gluon.

@becauseofAI any update on this?

Add Per Class Analysis and Class Confusion Matrix

Can the author approve this ?

Cannot generate Minimum Spanning Tree; the implemented Prim's does not produce the full minimum spanning tree .. cant get dbcv score /home/ec2-user/anaconda3/envs/pytorch_latest_p36/lib/python3.6/site-packages/hdbscan/hdbscan_.py:219: UserWarning: Cannot generate Minimum Spanning Tree; the implemented...

Sum of membership_vector not equal to 1

+1 checking in again to see if its fixed else adding a #featureRequest for it

How to extend the max-number of candles from 5k to 20k (allowed by premium subscribtion) ?

Did you manage to get 22k ?@galeese

Any plans for PRequesting in detectron2 repo?

It wont have the detectron2's open support for converting c2 to onnx/tensorrt. The way this model is implemented, the symbol cant be generated to productionize it. (Detectron2 has the same...

Commercial issue

@qdd1234 were you able to get the commercial license ?

The MAP of auther's pre-trained model EfficientDet-d0 on VOC is about 0.56?

> My model of efficient-d0 trained 255 epoches provides mAP of 74.8% on VOC 07, and 70.7% on VOC12 For testing on 2007 did you use 2007+2012 trainval for training...

Jay Desai