Felix Dangel
Felix Dangel
Hi, thanks for the nice themes! my question is similar to the remark on [changing the background of code blocks](https://github.com/fniessen/org-html-themes#change-the-background-code-blocks) in the `README`. But I would like to change the...
Problem: `cifar100_wrn404` forward pass is not working when adding regularization. This is caused by a missing return statement that creates the regularization groups. Demo: - Before: Forward pass without regularization...
The pooling indices in `MaxPool2d` are currently computed by performing a second forward pass, which can be avoided.
Hi, thanks for this great package! This is primarily a clarification question. I ran into this [comment](https://github.com/arogozhnikov/einops/blob/fcd36c9b017d52e452a301e75c1373b75ec23ee0/einops/einops.py#L833-L834) in `einops.einsum` which states that the index un-grouping syntax of `rearrange` (e.g. `'(a...
BackPACK's extensions that rely on the probabilistic interpretation of a loss function as a negative log likelihood (quantities based on the Fisher, i.e. `BatchDiagGGNMC`, `DiagGGNMC`, `SqrtGGNMC`, `KFAC`) are limited to...
Hi, I ran into a memory leak when using `pytaco` (I am using commit [`2b8ece4c230a5f0f0a74bc6f48e28edfb6c1c95e`](https://github.com/tensor-compiler/taco/tree/2b8ece4c230a5f0f0a74bc6f48e28edfb6c1c95e), the current `master` at the time of writing). The following script demonstrates the memory leak....
The implementation of `unfoldNd` relies on one-hot convolution. This means the convolution kernels are highly sparse. Hence, the code could run faster when using sparse tensors. Open questions: * What...
At the moment, all methods fall back to `.to_dense` and `.from_dense`.
At the moment, all methods fall back to `.to_dense` and `.from_dense`.
During one of our internal discussions, we realized that the code starts to accumulate multiple re-scaling operations which are required to avoid over/under-floating when using `float16` (e.g. using the average...