Jascha Sohl-Dickstein comments

Results 7 comments of


                                            Jascha Sohl-Dickstein

Why is the standard deviation always within [0, 1] and why do I get negative or NaN covariance values?

HI Amir-Arsalan. Thanks for your question! I wonder, are you sure the std/var values should be larger? Remember that the variance of a posterior is never larger than the variance...

Clarify that lax.top_k returns values and corresponding indices in descending order

Yes please! On Tue, Apr 16, 2024, 10:15 Peter Hawkins ***@***.***> wrote: > ***@***.**** commented on this pull request. > > This is explicitly not something we promise: there are...

Clarify that lax.top_k returns values and corresponding indices in descending order

I just made that change to the PR.

Clarify that lax.top_k returns values and corresponding indices in descending order

Done. PTAL.

data_path issue

Yeah, I also would strongly recommend building on something more modern. I would have had trouble debugging that error with an upstream library at the time I wrote the code...

Integrate LAHMC with PyMC3

I think this is a really good idea! Yes, I will do this. Question - does PyMC support running multiple particles in parallel (eg, with each column in the state...

Make sure the theano bits work on CPUs

hardcode to theano.config.floatX On Thu, Feb 4, 2016 at 8:36 PM Andrew Berger [email protected] wrote: > The theano implementation currently barfs when run on a CPU. I think this >...