orbit icon indicating copy to clipboard operation
orbit copied to clipboard

Stan backend doesn't response

Open JustM57 opened this issue 3 years ago • 2 comments

I'm using ubuntu 22.04 with python=3.10 and numpy 1.23.0 There was an issue installing pystan(I've managed to install it only from source (2.19.1.1).

The only model I can use is LGT as other don't pass C++ compilation.

JustM57 avatar Aug 05 '22 15:08 JustM57

Did you try following the installation guide of pystan2? Did you try python 3.8 instead? python 3.10 is not formally supported by orbit.

edwinnglabs avatar Aug 06 '22 04:08 edwinnglabs

I have similar issues. I've tried several approaches: When using Python 3.8 or 3.9, running pip install orbit-ml invokes legacy pystan compilation via setup.py as its wheel is not precompiled for Python 3.8+. This is not an issues for use on my local machine, but I need it to run on company cloud and there I got gcc: fatal error: Killed signal terminated program cc1plus compilation terminated. because it exceeds the memory limit.

When using Python 3.7, compilation is invoked when fitting the model: INFO:pystan:COMPILING THE C++ CODE FOR MODEL anon_model_aec9f19b676e0ecff1bcca082962e24a NOW. on cloud this is terminated due to time limit.

ondrejkopicka avatar Sep 19 '22 22:09 ondrejkopicka

I have similar issues. I've tried several approaches: When using Python 3.8 or 3.9, running pip install orbit-ml invokes legacy pystan compilation via setup.py as its wheel is not precompiled for Python 3.8+. This is not an issues for use on my local machine, but I need it to run on company cloud and there I got gcc: fatal error: Killed signal terminated program cc1plus compilation terminated. because it exceeds the memory limit.

When using Python 3.7, compilation is invoked when fitting the model: INFO:pystan:COMPILING THE C++ CODE FOR MODEL anon_model_aec9f19b676e0ecff1bcca082962e24a NOW. on cloud this is terminated due to time limit.

I have similar issue with memory limit. Pre compiling the stan script with orbit.utils.stan.compile_stan_model(<model_name>)" works for me.

Eve-Chen avatar Apr 07 '23 00:04 Eve-Chen

recommend to try latest version which uses cmdstanpy instead.

edwinnglabs avatar Dec 19 '23 07:12 edwinnglabs