Open jbueltemeier opened 2 years ago
Thanks a lot @jbueltemeier! Two things before I actually start to review:
chapters
directory. Given that we only use sections in this paper, please rename the directory.Thank you very much, @pmeier. If you have a good idea to improve this, feel free to push to this branch.
I'm unable to build the PDF. I think the whole graphics/*
folder is missing in this PR.
This is because the folder with the result images is currently not committed, as there are more than 150 image files in the folder and changes can still take place. So I wanted to integrate it at the end, but if you already need it to build the PDF I can do it now.
What size of the folder are we talking about? Can we maybe provide the scripts (if they differ from our replication scripts (which the shouldn't)) and only commit the *.tex
files that generate the figures? That way we could provide the images from our download server, e.g. https://download.pystiche.org/replication_paper/images.tar.gz
Currently the size of the folder is just over 45 MB.
Could you zip that and send it to me via email? I'll upload it afterwards.
Done. Could you push the graphics folder minus the images? I'll adjust the archive later so we can extract directly into it. On a side note: please store the images as .jpg
's. I'd wager a guess that this already brings down the archive size significantly.
I've cleaned up the branch quite a bit. Compilation should happen through the default command the template provides, i.e. make
. This currently fails (for me) and I've opened ReScience/template#25.
@jbueltemeier I've added the workaround from the issue above into the README.md
. A fresh clone of the repository now works, but our compilation still fails. Could you fix that?
Thanks a lot @pmeier for the workaround.
As far as I can see, the compilation fails because a few packages and commands (mainly for reference purposes) are missing, which I have added to article.tex. The missing packages are standalone, which could be removed, but also lscape for the tables in the appendix, which I think is necessary to display the tables better. So now I would add this file (article.tex from the template) back or do you see another possibility?
Can't we put the imports on top of content.tex
? If not, I'll send a fix to include the automatically in article.tex
. Just \usepackage{standalone}
and \usepackage{lscape}
?
No this is not possible, because content.tex
is inside the \begin{document}
statement and the packages have to be called before. Currently only the two packages (\usepackage{standalone}
and \usepackage{lscape}
) are needed, but more might be needed later.
Let me send a fix then.
@jbueltemeier I've added a preamble.tex
file in which you can put everything that needs to be loaded before \begin{document}
. For this file to be included in the document, you need to run download.py
again.
That being said, I'm still unable to build the document. The commands that I put in the README.md
are the ones suggested to me in the issue above and they indeed work for the bare template. If you are able to compile the article, which commands are you using?
I have added the missing commands. Currently I only get two error messages. The filename of schultenhof_mettingen_bauerngarten_8__j.-h._janßen.jpg
in images/gatys_et_al_2017/source/
seems to be incorrect after the download (ß is not correct) and the command \codeSWH
from metadata.tex
does not seem to work properly, perhaps because metadata.yaml
is not yet completely filled in. Are you getting the same errors or is it just my system?
I've re-uploaded the images you sent me to https://download.pystiche.org/replication-paper/images.tar.gz
Here is the first draft for a paper to be published in ReScience . It is based on the ReScience template.
It is currently being edited, feel free to comment on details to improve the paper.
Currently, individual chapters for the results, as well as the information on the hyperparameters used there, need to be attached.