Closed xendk closed 2 months ago
Nice, I'll have a closer look soon. I'm debating a bit whether to include those workarounds in the package, or as copy-and-paste examples in the wiki...
Also, regarding PyTorch, here's [one example])https://devdocs.io/pytorch/generated/torch.nn.crossentropyloss#torch.nn.CrossEntropyLoss). In Emacs it displays as
input has to be a Tensor of size either (minibatch,C)(minibatch, C) or
(minibatch,C,d1,d2,...,dK)(minibatch, C, d_1, d_2, ..., d_K) with K≥1K \geq 1 for the
K-dimensional case (described later).
I'm debating a bit whether to include those workarounds in the package, or as copy-and-paste examples in the wiki...
Well, as long it's a few style bug fixes, the best new user experience would be to include them.
one example
Hmm, both missing stuff and duplicated stuff. I'll look into it a day when I'm not as tired. But not being a math geek, what would be the most apropiate rendering (minibatch,C,d1,d2,...,dK)
or (minibatch, C, d_1, d_2, ..., d_K)
?
I've changed the defcustom
to defvar
.
After a bit of experimenting on pytorch, I've come up with this quick hack:
(defun devdocs--sphinx-tag-math (dom)
(when-let (semantics (dom-child-by-tag dom 'semantics))
(when-let (annotation (dom-child-by-tag semantics 'annotation))
(insert (dom-text annotation)))))
(setq devdocs-extra-rendering-functions '((sphinx (math . devdocs--sphinx-tag-math))))
It replaces the math
block with the TeX annotation
element. Requires that one is capable of reading TeX, but it's the best representation in the HTML. I don't know if there's any TeX packages that could be helpful with rendering.
Hi again, and sorry for the delay to react.
Note that you should actually write
(push '(a . xen-crystal-tag-a) (alist-get 'crystal devdocs-extra-rendering-functions))
otherwise you risk overwriting other special rendering functions. I will comment about it in the README.
Excellent.
Note that you should actually write
Thanks, I've adjusted my init file (in the process of replacing use-package with setup.el).
As requested in #37. This works for me, together with:
Haven't figured out the PyTorch issue yet, as I don't seem to be able to find an example.
Could use better documentation for the custom var, but I'm already flirting with the 15 line limit, and the paperwork is going to take some time.