Style the model cards and implement a broader set of data points

[sschafft]

Do a first pass at styling the PDF. Things to prove:

• [x] Fonts?
• [x] Colours?
• [x] Spacing?
• [x] Conditional rendering?

To do:

• [x] Make sure all data points are available in the PDF
• [x] Try to recreate an example to learn how to do styling
• [x] Meet with a designer
• [x] Get some initial designs
• [x] Style the output to match initial designs

[wayferer]

• Fonts: https://www.overleaf.com/learn/latex/Font_sizes%2C_families%2C_and_styles
• Colours: https://www.overleaf.com/learn/latex/Using_colors_in_LaTeX
• Spacing: https://www.overleaf.com/learn/latex/Line_breaks_and_blank_spaces
• Conditional rendering: See https://github.com/planecards/planecards-cli/pull/32

[wayferer]

@jlebensold should the properties for each section be hard coded in the template as they will be a fixed set or would it be better to loop through them in case they change in the future?

Section being Model Details / Intended Use etc Properties being refs / tco2e etc

[jlebensold]

I think we can assume that the properties of the plane card will be fixed for now. The template doesn't need to be that smart. However if iterating over "other details" or something of the sort makes sense, then we can do that. We will need a way to show all the binary facts about the model somewhere, but I would say that this isn't immediately critical and probably is the most likely part of the spec to change. Focus on the model details / model card sections for now.

[wayferer]

Thanks Jon. I'm having a few issues rendering out the properties from the YAML like the following. I tried searching for how to do it in Latex but couldn't get it working so that it would render each as an "item":

intended_use:
    - Llama 2 is intended for commercial and research use in English.
    - Tuned models are intended for assistant-like chat.
    - Pretrained models can be adapted for a variety of natural language generation
      tasks.
    - Developers may fine-tune Llama 2 models for languages beyond English provided
      they comply with the Llama 2 Community License and the Acceptable Use Policy.
    - Use in any manner that violates applicable laws or regulations is out-of-scope.

i.e.

• Llama 2 is intended for commercial and research use in English.
• Tuned models are intended for assistant-like chat.
• Pretrained models can be adapted for a variety of natural language generation tasks.
• Developers may fine-tune Llama 2 models for languages beyond English provided they comply with the Llama 2 Community License and the Acceptable Use Policy.
• Use in any manner that violates applicable laws or regulations is out-of-scope.

I tried forloops and forcsvlist but couldn't quite get there

[wayferer]

Todo

• [ ] Add an example of each of the components from the design (this should just be basic i.e. don't fine tune)
  • [x] Header
  • [x] Version of builder
  • [x] Page layout
  • [x] Section title + rule + descriptor
  • [x] Section table
  • [x] Metric bar
  • [ ] Link styling
  • [ ] Sub heading styling
  • [ ] Paragraph styling
  • [ ] List styling
  • [x] Boxed section styling
• [ ] Fine tune each of the components
  • [ ] Header
  • [ ] Font
  • [ ] Spacing
  • [ ] Alignment
  • [ ] Page layout
  • [ ] Left column should be 60%
  • [ ] Section table
  • [ ] Side by side equal columns
  • [ ] Render ticks in place of 1
  • [ ] Section title
  • [ ] Font style
  • [ ] Spacing
  • [ ] Rule style
  • [ ] Reusable component
  • [ ] Override for when in a boxed section