GSoC #10: All Good Things…
Published:
This will be my final blogpost as part of my GSoC series which can be found here.
Blog posts
2025
GSoC #9: The Final Push
Published:
Although I still have two weeks left, I’ll be at the ODYSSEY Alignment Conference next week so I’ll have limited time, which means I’ll be leading the push this week to get the main points sorted.
GSoC #8: A Light at the End of the Tunnel
Published:
With the final product due about three weeks from now, I have been busy tidying things up so we have an end-to-end solution ready, with a few self-contained issues/TODOs remaining, which can be wrapped up into seperate issues quite readily. As of yesterday, I’ve gotten INLA working end-to-end with pmx.fit, as well as having added a few test cases and some guardrails (namely that the latent must be a Gaussian in precision form).
GSoC #7: Taking Stock
Published:
These past two weeks were, like the one before it, largely about dealing with the implementation in code. In theory, the marginalization routine works from top to bottom, however there are a few outstanding bugs that need to be squashed.
GSoC #6: Less Math More Code
Published:
This week was all about trying to wrangle PyTensor to correctly marginalise out the hyperparameters whilst leaving \(x\) untouched by pm.sample. In contrast to prior weeks, there was much less focus on the theory, as this time it was a matter of trying to get PyMC and PyTensor to do what I want.
GSoC #5: Putting it all together
Published:
This week I got started on my next PR to implement “regular” (i.e. non-Laplace) marginalisation on the hyperparameters, and also develop a skeleton for a user-facing fit_INLA method which would essentially be the final product.
GSoC #4: Spoilt for Choice
Published:
This week I was finally able to get my first INLA-related PR merged in!
GSoC #3: A Slight Detour
Published:
My PR’s almost done - I promise!!
GSoC #2: \(x_0\) marks the spot on the MAP
Published:
Over this past week, I’ve been working to get this issue closed out. INLA allows us to perform Bayesian inference over models with latent Gaussians. By “latent”, we mean that the Gaussian component is not what we directly observe, but is related to the observed data nonetheless.
GSoC #1: Welcome!
Published:
Hi there! This is the first in my blog series documenting my Google Summer of Code 2025 project. Every week (or fortnight, depending on my time), I’ll be posting a brief update to document my progress and share my thoughts on my work.