Skip to content

ExceptionRegret/gitcoin-deep-funding-level1-model

Repository files navigation

Gitcoin Deep Funding Level I Model

Sponsor

Level I submission package and reproducible modeling code for the Gitcoin x Ethereum Foundation Deep Funding Contest.

Project Scope

This repository contains the Level I model only:

  • predict weights for the Level I repos relative to ethereum
  • generate a valid submission CSV
  • include the writeup required for prize eligibility

Final Submission

Primary submission file:

  • level1/submission_level1_recommended.csv

Current selected model:

  • compact pairwise model using:
    • l1-predictions.csv
    • originality-weights.csv
    • Level III outgoing dependency count

Repository Layout

  • level1/
    • Level I scripts, outputs, notes, and writeup
  • originality-predictions.csv
    • contest-source Level II prior
  • originality-weights.csv
    • contest-source Level II prior
  • pairs_to_predict.csv
    • contest-source Level III dependency graph
  • seedReposWithDependenciesAndWeights.json
    • contest-source Level III weighted dependency prior

Main Files

  • level1/generate_level1_submission.py
  • level1/validate_level1_submission.py
  • level1/analyze_level1_trial.py
  • level1/level1_writeup.docx
  • level1/level1_writeup.md
  • level1/submission_level1_recommended.csv

Requirements

Python packages used:

  • numpy
  • scipy
  • python-docx

Install them with:

python -m pip install -r requirements.txt

Reproduce The Final CSV

From the repository root:

python .\level1\generate_level1_submission.py --method pairwise_simple_model --pairwise-alpha 20 --output submission_level1_recommended.csv
python .\level1\validate_level1_submission.py --submission submission_level1_recommended.csv

Writeup

Prize eligibility requires a writeup. This repo includes:

  • level1/level1_writeup.docx
  • level1/level1_writeup.md

Notes

  • This repository is intentionally focused on Level I only.
  • The model was selected using local historical calibration proxies.
  • Hidden contest jury data is not included here, so local proxy performance is not a guarantee of leaderboard score.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages