https://github.com/eykrehbein/llm-sheets

A spreadsheet that fills itself
https://github.com/eykrehbein/llm-sheets

Last synced: about 1 month ago
JSON representation

A spreadsheet that fills itself

• Host: GitHub
• URL: https://github.com/eykrehbein/llm-sheets
• Owner: eykrehbein
• Created: 2024-08-03T19:47:10.000Z (5 months ago)
• Default Branch: main
• Last Pushed: 2024-08-05T15:35:25.000Z (5 months ago)
• Last Synced: 2024-08-05T19:33:25.387Z (5 months ago)
• Language: TypeScript
• Homepage:
• Size: 675 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# LLMSheets

> A spreadsheet that fills itself.

![](./docs/demo.gif)

## WIP

This is a WIP of a hackathon project. 

What's currently possible:

- Creating spreadsheets

- Creating columns with certain types and descriptions, that will be filled by LLMs

- (locally) uploading PDF files (with up to 20 pages) that are split into pages and passed to LLMs as images for extraction

- Streaming object responses to the table

Note:

- The code is not cleaned up – some things might be pretty hacky

- Currently Python is needed to split the uploaded PDFs into pages and convert those pages into images (couldn't find a stable way to do this with Node)

- All data is stored in the browser using IndexedDB

- WIP

## Running the project locally

1. Clone the repo

2. Run `pnpm i`

3. Create a python venv called `venv` in `apps/web` and install the packages `pdf2image` and `PyPDF2` and make sure [poppler](https://poppler.freedesktop.org/) is installed

4. Currently, `claude-3.5-sonnet` is used as LLM. Make sure `ANTHROPIC_API_KEY` is set in `apps/web/.env` 

5. Run the next project using `pnpm dev` or `pnpm build` & `pnpm start`