Sam Shleifer

Also the parrelization fail warnings are 13 seconds apart but cite timeout limits that are 1200/2400/4800 seconds apart?

thanks that helped me figure it out! On Sun, Feb 11, 2018 at 3:08 AM anttttti wrote: > Usually this is something wrong with the input. Try adding method="serial" >...

yes, I think my loss got stable after roughly 12h training on 1 GPU. On Wed, May 30, 2018 at 3:53 AM ShiAGou wrote: > First of all, thank you...

From here, looks to me like you are evaluating loss on a per image basis, which is not a very good accurate proxy for your train loss over the whole...

Dont know windows well, but i would guess problem is with PYTHONPATH What is your pythonpath, where are you running code from?

Hmm. can you paste a traceback? On Wed, Jan 24, 2018 at 5:13 PM zeynali wrote: > no i switch to ubuntu 16.04 , i running this code from object_detection...

Yeah! On Sat, Feb 10, 2018 at 7:42 AM zeynali wrote: > is this correct ? > > mm@mm:~/object_detection_kitti/object_detection$ python3 create_dataset.py > > > Traceback (most recent call last): >...

what does echo $PATH in terminal produce? On Sun, Feb 11, 2018 at 12:24 AM zeynali wrote: > So what do i do then؟ > > — > You are...

Good catch! If you `git pull` create_dataset should create train.txt as a side effect, I believe

you can either run create_dataset again or you can read the code On Mon 9. Apr 2018 at 20:52, ghimiremukti wrote: > Hi @sshleifer , thanks for the reply. Now...