cereusperuv

@paint-it-black You have probably moved on from this problem a long time ago, but I just found your post and wanted to share my thoughts on a similar problem I've...

Hi @CamDavidsonPilon and thanks for your prompt reply. I will send you my dataset shortly, just have to edit it a tad first. Meanwhile, I can tell you that 1....

Grouping leads to a never ending stream of new autograd warning messages of the type ``` C:\Tools\Anaconda3\envs\TestEnv\lib\site-packages\autograd\tracer.py:48: RuntimeWarning: overflow encountered in power return f_raw(*args, **kwargs) C:\Tools\Anaconda3\envs\TestEnv\lib\site-packages\autograd\tracer.py:48: RuntimeWarning: invalid value encountered...

Grouping and penalizing (0.001) makes it work without any warnings, but the results are non-sensical. It also takes forever to fit (didn't time it, but something like 20 minutes on...

@CamDavidsonPilon Sorry for taking so long to follow up on this, I have been busy with other work. There seems to be stability issues with the BG-BB autograd optimizer. For...

@pallenmar Sorry for my inactivity, have been working on other projects. Today I tried moving the ab/gd-log-betas inside the for loop like you suggested but I saw no change in...

@Trollgeir Still unresolved. Not working actively on it anymore.