A CLI for Trustworthy Pre-Trained Model Re-Use

Setup

LLM

If you are using Gemini-2.0-flash, place your API key into gemini_key.txt or in the environment variable called GEMINI_API_KEY in the .env file. If you are using Purdue GenAI Studio (preferred), then place your API key into the environment variable called GEN_AI_STUDIO_API_KEY in the .env file.

Note: .env file must be created locally for each user

GitHub

Optionally, set up a GitHub token for higher rate limits:

• Preferred: set an environment variable GITHUB_TOKEN
• Or create a file git_token.txt with your token in the first line

HuggingFace

Optionally, set up a Hugging Face token for higher rate limits:

• Preferred: set an environment variable HF_TOKEN
• Or create a file hf_key.txt with your token on the first line (this file is ignored by git)
• To avoid OS permission issues with Hugging Face, set a project-local cache

Usage

Run ./run install to install dependencies or ./run URL_FILE where URL_FILE is filepath to list of URLs

Testing URL Parsing

To test the URL parsing functionality, create a text file with CSV format:

<code_link_1>, <dataset_link_1>,<model_link_1>
<code_link_2>, <dataset_link_2>,<model_link_2>
... and so on

Then run: ./run your_test_file.txt Note: I just used the sample_input.txt from Piazza

Contributors

• Mikhail Golovenchits
• Vatsal Dudhaiya
• Murad Ibrahimov
• Jake Scherer