Iain Murray
@imurray.bsky.social
Professor of Machine Learning and Inference, Edinburgh Informatics, Formerly Amazon Scholar. Opinions are my own. Also https://homepages.inf.ed.ac.uk/imurray2/ and https://mastodon.social/@imurray and https://x.com/driainmurray
The 13x13 grid became 14x14. It's really hard to get image models to make a board with the correct number of evenly spaced lines and star points. I'd thought it might copy a reference board into a scene, but it can't even do a simple copy, showing limits to guidance that's possible with this model.
May 3, 2025 at 10:25 AM
The 13x13 grid became 14x14. It's really hard to get image models to make a board with the correct number of evenly spaced lines and star points. I'd thought it might copy a reference board into a scene, but it can't even do a simple copy, showing limits to guidance that's possible with this model.
Standard normalizing flows are "approximate" too: e.g., standard implementations are often not exactly invertible in practice. But ODE solvers can use adaptive computations to control the errors. So they're not intractable in the same way that (say) a Boltzmann machine is.
November 29, 2024 at 5:00 PM
Standard normalizing flows are "approximate" too: e.g., standard implementations are often not exactly invertible in practice. But ODE solvers can use adaptive computations to control the errors. So they're not intractable in the same way that (say) a Boltzmann machine is.
November 20, 2024 at 8:00 PM
I do ML + Bayesian inference.
November 18, 2024 at 6:15 PM
I do ML + Bayesian inference.