Skip to content

Conversation

@xukai92
Copy link
Member

@xukai92 xukai92 commented Apr 11, 2017

Run the normal mixture model in Stan.

The script seems to work.

NOTE: I guess the build filed written by Stan.jl team is problematic itself. We don't necessary need to "build Stan.jl successfully" to use Stan.jl. From the log we can see the Stan model is successfully built and run, even we didn't build the Stan.jl successfully. This happens to my local Linux machine as well.

@yebai
Copy link
Member

yebai commented Apr 11, 2017

Do we actually need to build Stan.jl?

@xukai92
Copy link
Member Author

xukai92 commented Apr 11, 2017

We don't need to build it explicitly because when we call Pkg.add("Stan") it actually builds Stan.jl. I will remove Pkg.build("Stan").

@xukai92
Copy link
Member Author

xukai92 commented Apr 11, 2017

I think the script for bench works now. I'm gonna add Stan version of bernoulli and simplegauss now.

@yebai
Copy link
Member

yebai commented Apr 11, 2017

It might be useful to compute and print the difference (e.g. mean, variance) between Stan and Turing

@xukai92
Copy link
Member Author

xukai92 commented Apr 11, 2017

Yeah. That makes a lot of sense. Will do!

@xukai92 xukai92 requested a review from yebai April 11, 2017 19:25
@yebai yebai merged commit c64a725 into master Apr 11, 2017
@yebai yebai deleted the run-stan-models branch April 11, 2017 20:39
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants