Unbundled front-end - Githubissues

microsoft / lida

Automatic Generation of Visualizations and Infographics using Large Language Models

https://microsoft.github.io/lida/

MIT License

2.66k stars 282 forks source link

Unbundled front-end #18

Closed 0xaaiden closed 11 months ago

0xaaiden commented 1 year ago

Is it possible to publish the frontend ui currently used for Lida? Thanks!

victordibia commented 12 months ago

Hi @0xaaiden ,

At this time, the core OSS focus is underlying apis with the UI being optional. Currently no plans to release the UI at the moment.

-V.

benheymink commented 12 months ago

@victordibia Could you at least make those sample files available? It would be interesting to see how the examples are being converted!

victordibia commented 12 months ago

Hi @benheymink ,

Thanks for the note. Can you give some examples of what you mean by "being converted"

OTOMH, I can think of the following:

examples of how the goals are parsed etc?
how images can be rendered in say an tag?

Are these in line with what you are thinking?

In general, a good place to start in terms of understanding how to build front ends with lida is to run lida ui --port 8000 --docs and then visit localhost:8000/api/docs to view the api specification for each type of action.

I'll work on adding some documentation on this.

benheymink commented 12 months ago

@victordibia Sure, sorry - I should have been more clearer. I wanted to quickly spin up the UI to evaluate the project. Couldn't get it running locally (because of some of the same issues reported in this repo (#25 , #19 etc.).

Ended up installing the package in a replit sandbox online, but first thing I wanted to do is take a look at the sample files the UI presents to try - I wanted to understand the exact markup of those CSV files and JSON files, but couldn't see anywhere in the app to actually view the 'raw' data, if you see what I mean.

benheymink commented 12 months ago

Basically the contents of these files @victordibia

victordibia commented 12 months ago

Ah .. got it. The dataset files in the screenshot are actually direct links from the vega_data github repo. For example this is a link to stocks.

Also, the python tutorial notebook has an example of passing a dataset url.

victordibia commented 12 months ago

Thanks!

Also the basemap installation issue #19 is addressed by making basemap optional. e.g.pip install lida[tools].
So going forward, basemap should not be required. (basemap is only needed when lida suggests visualization code that includes maps).

pip install lida==0.0.8