course-content VS score_sde

Yes. And if you want to get a jump start on what a computational neuroscience PhD might look like, check out our materials at https://compneuro.neuromatch.io. Like a PhD in a box... and all free to use/explore!

The neuromatch computational neuroscience course also seems quite interesting, though maybe less of practical use.

https://compneuro.neuromatch.io/

Recent research like "Relating transformers to models and neural representations of the hippocampal formation" might make it more relevant though (https://arxiv.org/abs/2112.04035v2)

quote from the abstract of that paper:

check out the curated datasets and project outlines for Neuromatch Academy's computational neuroscience course for some inspiration! https://compneuro.neuromatch.io

One possible thing I saw is taking https://compneuro.neuromatch.io/ courses, Although I'm not sure of the effect it adds to my CV.

https://github.com/NeuromatchAcademy/course-content https://github.com/NeuromatchAcademy/course-content-dl

If you'd like to learn more about it, you can check out last year's Comp Neuro course contents here, last year's Deep Learning course contents here, read the paper we wrote about the original NMA here, read our Nature editorial, or the Lancet article00074-0/fulltext) about us.

Check out Neuromatch Academy. If you don't have time or inclination to do the summer school this summer, you can simply access all the videos and tutorials for free at https://compneuro.neuromatch.io

https://compneuro.neuromatch.io has all the materials for free, including lecture videos and hands on Google Colab notebook programming exercises. You can also apply to take the course synchronously with a "pod" of 10 other people in a 3-week summer intensive online, with a TA to guide you and group projects to cement your understanding. To apply, check out http://academy.neuromatch.io. Applications open soon and it sounds like you'd be a great fit.

Yes, Neuromatch Academy. Their whole course is available as a Jupyter Book here

Maybe better than any single book or paper, I would recommend you check out neuromatch academy. There are 2 courses, computational neuroscience and deep learning and everything is free and open source. Play with the code, listen to the lectures, even do a "canned project" if you want experience.

Perhaps the ODE interpretation would be helpful (see here and here) which turns DDPMs into neural ODEs using the Fokker-Planck equation so after the initial starting noise, the sampling process is deterministic. If samples are noisy even with the full number of steps then you might need to increase the number of steps further.

Turns out that diffusion models also define a certain differential equation, making it a neural ODE. Then you can just integrate the ODE in the other direction to get the exact inverse for the DDPM (it's not entirely exact b/c of numerical error in the solver, but close enough)

This new approach to generative modelling looks very intriguing.

In a similar ilk, there's this ICLR paper from this year using stochastic differential equations for generative modelling: https://arxiv.org/abs/2011.13456