Work in Progress: Review of Literature on Growing Neural Networks

This repository serves as a place for keeping personal notes on relevant literature around growing artificial neural network.

The article aims to summarize the existing literature concerned with growing artificial neural networks. For each paper it will list the most significant contribution.

Contributing

Structure of the Content

All articles in src/articles.bib which have a matching summary file (src/summaries/{bibkey}.md) are included in the list of articles (ordered by year and title).

To add a new article, simply add it to the bib file and add a new summary file using the same key. The rest is handled by the github actions.

General comments, as well as the list of contributors, can be found in src/intro.md.

Please verify, the build is handled correctly before pushing/merging to main (this includes the html as well as the pdf file).

Images

The best way to include images currently seems to be to place the image as an svg as well as an pdf in the src/static/img directory and then use the usual markdown syntax, leaving out the fle extension. The Makefile is configured such that for a PDF output the PDF image is used, for the html file, svgs are used.

![Description](img/<image file name>)

Alternatively, one can use PNG images (include the extension in the file name).

Width or height can be adjusted using curly braces:

![Description](img/<image file name>){ widht=10cm }

Local Building and Testing

Content pushed/merged to the main branch is automatically deployed.

Locally, distribution files can be generated using the Makefile. A python3 installation with the packages listed in requirements.txt is required to build article summaries into one file.

Additionally, pandoc as well as latex installation is required.

make pdf: Produces the PDF document (which can the be found in the dist directory).

make html: Produces the html file, copies the stylesheet and references file, and also produces the PDF (as it can be accessed via the web page).

make dev: Runs a local http server in the dist folde (after creating the required files, see make html). The build webpage can then be accessed via localhost:8000.

Alternatively, use the provided Dockerfile, mount this repository to /data and run one of the make commands as such:

docker run -v $(pwd):/data ghcr.io/plonerma/pandoc-latex-python3 make html