These are chat archives for beniz/deepdetect

20th
May 2017
alkollo
@alkollo
May 20 2017 05:35
hello, I tryed the training image classifier tutorial, all went fine until the first test. At the test time (after xx iterations) dede return "MemoryData does not transform array data on Reset()" . I have googled a bit and found that it is just a warning but dede get stuck on it and seems to never pass the test to continue iterations. I use the CPU version of dede, have 8 GB of RAM that is never full, a small batch size of 32. Thanks by advance to anyone that have a clue on this. I also opened an issue with all logs beniz/deepdetect#316
roysG
@roysG
May 20 2017 07:46
I want to run the "deepdetect" in lambda amazon, can i do it, does it possible?
Emmanuel Benazera
@beniz
May 20 2017 08:11
@sunatthegilddotcom you may want to try, yes
@alkollo this is simply a warning during test forward passes, it does not affect training or the results. As a remainder, training on CPU is ~50x slower than on GPU.
@roysG No idea about lambda nor have I heard about using DD+lambda. We do maintain official AMI on AWS, or you can build your own.
alkollo
@alkollo
May 20 2017 09:59
@beniz . I was not sure to 100% it was a bug or not, even by reading some docs about it. I let the service run all night and you're right, it continues to iterate. CPU is far slower than GPU sure but it took a lot of time beetwen test and following iteration. I'm waiting my AMI instance at Amazon to try on GPU. Thanks a lot !