Closed KaelanDt closed 2 months ago
I don't feel strongly on this, the
sample_identity_diffusion
function is already kinda not public facing since it only lives inthermox.sampler.sample_identity_diffusion
unlikethermox.sample
.Happy to add the underscore if we like but if we do we should also add it to
thermox.log_prob.log_prob_identity_diffusion
to be consistent.
That's true, let's keep it that way. Thinking about it again, I just don't like that sample_identity_diffusion
can throw an error if you want to use it on a GPU, so what do you think of simply adding A_spd
as a kwarg to it?
I don't feel strongly on this, the
sample_identity_diffusion
function is already kinda not public facing since it only lives inthermox.sampler.sample_identity_diffusion
unlikethermox.sample
. Happy to add the underscore if we like but if we do we should also add it tothermox.log_prob.log_prob_identity_diffusion
to be consistent.That's true, let's keep it that way. Thinking about it again, I just don't like that
sample_identity_diffusion
can throw an error if you want to use it on a GPU, so what do you think of simply addingA_spd
as a kwarg to it?
Yes absolutely!! Same wavelength, I'd even made an issue for it #9 😆
we should also do it for thermox.log_prob.log_prob_identity_diffusion
Can we unify the docstring for A_spd
for log_prob
too?
I don't feel strongly on this, the
sample_identity_diffusion
function is already kinda not public facing since it only lives inthermox.sampler.sample_identity_diffusion
unlikethermox.sample
.Happy to add the underscore if we like but if we do we should also add it to
thermox.log_prob.log_prob_identity_diffusion
to be consistent.