einops VS tiny-cuda-nn

Not sure if the wrapper you’re talking about is your own custom code, but I really like using einops lately. It’s got similar axis naming capabilities and it dispatches to both numpy and pytorch

http://einops.rocks/

Thanks for the pointer. I can believe that a language that looks so different will find that different patterns and primitives are natural for it.

My experience from writing a lot of array-based code in NumPy/Matlab is that broadcasting absolutely has made it easier to write my code in those ecosystems. Axes of length 1 have often been in the right places already, or have been easy to insert. It's of course possible to create a big mess in any language; it seems likely that the NumPy code you saw could have been neater too.

In machine learning there can be many array dimensions floating around: batch-dims, sequence and/or channel-dims, weight matrices, and so on. It can be necessary to expand two or more dimensions, and/or line up dimensions quite carefully. Einops[1] has emerged from that community as a tool to succinctly express many operations that involve lots of array dimensions. You're likely to bump into more and more people who've used it, and again it seems there's some overlap with what Rank does. (And again, you'll see uses of Einops in the wild that are unnecessarily convoluted.)

[1] https://einops.rocks/ -- It works with all of the existing major array-based frameworks for Python (NumPy/PyTorch/Jax/etc), and the emerging array API standard for Python.

this is definitely what you're looking for: https://github.com/arogozhnikov/einops

Einops all the things! https://einops.rocks/

It depends on what are their main purposes. I know some figures who have done an amazing job in this field but never because of publications, e.g. https://github.com/lucidrains and https://github.com/rwightman, https://einops.rocks/

On tsalib's warp: this is very similar to einops. I think it might even be slightly more general. However, I'm honestly not sure to what extent tsalib is still maintained, as it looks like the most recent commit was about two years ago. (Which is a shame.)

Also see Einops: https://github.com/arogozhnikov/einops, which uses a einsum-like notation for various tensor operations used in deep learning.

https://einops.rocks/pytorch-examples.html shows how it can be used to implement various neural network architectures in a more simplified manor.

This combined with something like einops [1] ( intuitive reshaping library) can be a huge time saver.

[1] https://github.com/arogozhnikov/einops

In the opposite direction from your question is a very interesting project, TinyNN all implemented as close to the metal as possible and very fast: https://github.com/NVlabs/tiny-cuda-nn

This repo only implemented the rendering part of the NGP but is more simple and has a lesser amount of code compared to the original (Instant-NGP and tiny-cuda-nn).

...a C++ library with a CUDA backend. But these high-performance building blocks might only be saturating the GPU fully if the data is large enough.

I haven't looked at implementing these things, but I imagine uf you have smaller networks and thus less data, the large building blocks may not be optimal. You may for example want to fuse some operations to reduce memory latency from repeated memory access.

In PyTorch world, there are approaches for small networks as well, there is https://github.com/NVlabs/tiny-cuda-nn - as far as I understand from the first link in the README, it makes clever use of the CUDA shared memory, which can hold all the weights of a tiny network (but not larger ones).