Philip Meier

Results 152 issues of Philip Meier

https://github.com/pystiche/pystiche/blob/71217c24557dfba05da5795547bf6f3034e7c66f/pystiche/enc/prepostprocessing.py#L11-L19 need to be put into https://github.com/pystiche/pystiche/blob/main/docs/source/api/pystiche.enc.rst

The CLI tests are slow, because almost all of them instantiate an MLE although it would be fine to do it only once for the whole module, since we don't...

After the merger of `pystiche.ops` and `pystiche.loss` it was forgotten to update the references in the documentation. For example: https://github.com/pystiche/pystiche/blob/main/docs/source/gist/loss.rst Basically all references of `pystiche.ops.*Operator` need to be changed into...

documentation

This reverts commit 30039013fb452a5abed7dd4ff637c06d0ae98960.

Revert "temporarily disable latex doc builds (#355)". This reverts commit 30039013fb452a5abed7dd4ff637c06d0ae98960.

do NOT merge

A long long time ago, I was fed up that PyTorch didn't provide an solution for testing two tensors for approximate equality. To overcome this I started [`pytorch_testing_utils`](https://github.com/pmeier/pytorch_testing_utils), which is...

enhancement
Good first issue

We are currently testing multiple `torch` / `torchvision` pairs in CI. To inject the pair we are using `tox`' `--force-dep` flag: https://github.com/pystiche/pystiche/blob/65f4d787e44b1ffbf7e5b6e48298ed8c7460e5a9/.github/workflows/tests.yml#L96-L101 Since `tox` is all about having separate virtual...

To be able to test the `pystiche.optim` convenience functions, we added pickled assets in #163. Unfortunately, they are unnecessary strict. This was hit before in #373 where we needed to...

enhancement

Right now if we want to access the names of the images stored in a `ImageCollection` we have to resort to something along the lines of ```python import pystiche_papers.gatys_ecker_bethge_2016 as...

enhancement

Right now `VGGEncoder` only support the convolutional stage https://github.com/pmeier/pystiche/blob/6ced94ad32d50e6dc7e69c7a0b5a57580c1df49b/pystiche/encoding/vgg.py#L45-L51 We should convert the subsequent `nn.Linear` layers into 1x1 `nn.Conv2d` and include them. This increases the memory requirement at creation, but...

enhancement