Someone
Apparently this is also an issue for unfree tests: https://logs.ofborg.org/?key=nixos/nixpkgs.256230&attempt_id=15fe1427-f73b-4e06-a084-11cd77b929e4
I'm not familiar with WSL, but from a brief search they seem to be deploying their `libcuda.so` in `/usr/lib/wsl/lib`: - https://discourse.nixos.org/t/pytorch-cuda-on-wsl/18267/9 - https://github.com/hasktorch/hasktorch/issues/535 - https://github.com/msteen/nixos-vscode-server/pull/44 - https://github.com/nix-community/NixOS-WSL/pull/221 Note that if...
Yes, you'd only need to symlink the libraries that version lock the driver: libcuda.so, libnvidia-ml.so... Adding that location into LD_LIBRARY_PATH should work, in principle. I haven'theard of LL_CUDA... before, is...
As in #52, we could just _agree_ to make it the default implementation of `retr` defined in the base class. This still leaves the problem of uncountably many possible `retr`-s...
> What is confusing there? Like, what do these methods do? I do know already `projx` stands for orthogonal projection of a vector from ambient space unto the manifold, but...
@calincru we've been thinking about providing implementations for `transport_follow_retr` in the base class, which would've defaulted to `retr` followed by `transp` as in the snippet you just referenced; there is,...
A year later, I'd say single repo and `.nn` sounds OK-ish I would be OK re-joining the effort somewhat after Nov 17
Great, here's first-Sunday-after-Nov-17
Maybe this could be generalized to `cargo.toml` and probably others: https://github.com/huggingface/safetensors