training_policies icon indicating copy to clipboard operation
training_policies copied to clipboard

Mask-RCNN clarifiation around num_image_candidates

Open jonathan-cohen-nvidia opened this issue 4 years ago • 3 comments

There was a rule previously that you could choose to do either:

Process the whole batch at a time and generate the total required number of patches (1000 * batches per chip)

OR

one image at a time, with exactly 1000 patches per image

Current HP Table (https://docs.google.com/spreadsheets/d/128r_G1KEE0C1PzuTnBGybeNsT-mXeORZhmWnqDgu8Ok/edit#gid=540034374 row 30 in the "Approved HP" tab) omits this detail. We should probably codify it somewhere.

@bitfort

jonathan-cohen-nvidia avatar Jun 05 '20 02:06 jonathan-cohen-nvidia

SWG:

Will follow up over email. Who wants 2000?

bitfort avatar Jun 11 '20 17:06 bitfort

AI(Victor) Email

bitfort avatar Jun 11 '20 17:06 bitfort

SWG:

This has been resolved in the HP table. Do what the table says.

bitfort avatar Jun 22 '20 17:06 bitfort