Ben Thompson

More good sources on "hessian-free" second order optimization methods. https://andrew.gibiansky.com/blog/machine-learning/gauss-newton-matrix/ http://www.cs.toronto.edu/~jmartens/docs/Deep_HessianFree.pdf

Not a priority, but Liz should do this because she read the book.

@MarcAntoineSchmidtQC and I talked about this briefly. My simple argument for keeping it was that it would allow us to keep having a conda-build step in the CI pipeline which...

The result of https://github.com/Quantco/quantcore.glm/pull/424 makes me think that this issue might have a substantial impact in certain narrow cases if someone ever wants to implement it. This will have a...

I'm not the expert here but I think there are issues here not captured by #5506. I'm sure fixing #5506 would improve the situation here. But, the example pyproject.toml I...

> I have a workaround for exactly this :) > https://github.com/jorenham/jax_pep503 Thanks! For other folks coming by, what jorenham has done is super useful for having a PyPI-compatible JAX package...

Oh well, I was hopeful for @jorenham 's solution, but it still turns out to be thwarted by problems with poetry 1.3.2 dependency handling. Perhaps the simplest reproducing example is...

> What if you explicitly exclude the versions you don't want, i.e. replacing `version = "0.4.4"` with `version = "==0.4.4,!=0.4.4+cuda11.cudnn86"`? Nice idea, thanks! The resulting poetry lock is unchanged. But,...

@dimbleby Out of curiosity, how should JAX handle this CUDA/no-CUDA situation in its releases? I am happy to file an issue with JAX.

It also seems feasible for Poetry to add something like a “precise” flag that ensures that `version = "0.4.4+cuda11.cudnn86"` is not used to satisfy a `version = "0.4.4", precise =...