🏗 GovernGPT

This project aims to develop GovernGPT which assists governors and professional delegates in DAOs to build and summarise proposals. This proposal is specifically geared to support DAOStewards a meta-governance group from BanklessDAO with active delegations in Euler.Finance, SAFE DAO & 1inchDAO.

You can read more about the project here.

🤖 Algovera Flow

This workflow will soon be added to the Algovera platform. Check it out if you're interested in using this workflow (and others) to augment your daily work with LLMs.

👪 Grants

You can get funded to build your LLM workflow through Algovera Grants. Each month, we give out $10k of grants to projects. For examples of previous proposals, check out our forum. For more information on the proposal process, check out our docs. We're very grateful to Ocean Protocol Foundation for sponsoring Algovera Grants.

Website | Twitter | Discord

🚜 Files

• scraper.rb: Scraper that scrapes over the governance proposals in Bankless DAO
• data_cleaner.rb: Data Cleaner that cleans out the unneccessary HTML tags from the scraped governance proposals
• prepare_dataset.rb: Convert HTML into the structured JSON format that contains title, headings, and sections. This JSON is used in turn to generate prompt/output pairs for OpenAI GPT model fine-tuning
• prepare_prompts.rb: Script that converts the scraped data in JSON format into the prompts/outputs used to fine-tune GPT.

📁 Data

• scraped_data/prompts_prepared.jsonl: Prompts and outputs prepared for fine-tuning after running openai tools fine_tunes.prepare_data -f gpt_data/prompts.jsonl script
• scraped_data/prompts.jsonl: Prompts and outputs initially prepared for fine-tuning
• scraped_data/json: Cleaned up JSON of governance proposals that contains title, headings, and sections. Used to generate prompts/outputs for fine-tuning
• scraped_data/clean_html: Cleaned up HTML of governance proposals
• scraped_data/html: Original raw HTML data scraped
• scraped_data/txt: Original raw TXT data scraped