https://github.com/jamesvorder/python-tddpp

This LLM generates code based on tests, and makes sure they pass.
https://github.com/jamesvorder/python-tddpp

automatic-program-repair automatic-programming llama llm tdd

Last synced: about 1 month ago
JSON representation

This LLM generates code based on tests, and makes sure they pass.

• Host: GitHub
• URL: https://github.com/jamesvorder/python-tddpp
• Owner: JamesVorder
• Created: 2024-07-12T13:11:41.000Z (7 months ago)
• Default Branch: main
• Last Pushed: 2024-08-06T17:56:00.000Z (6 months ago)
• Last Synced: 2024-12-10T15:35:43.504Z (about 2 months ago)
• Topics: automatic-program-repair, automatic-programming, llama, llm, tdd
• Language: Python
• Homepage:
• Size: 125 KB
• Stars: 1
• Watchers: 2
• Forks: 1
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# TDD++

The goal of this tool is to bring TDD to the next level.

Rather than writing tests first, let's write tests _only_.

Provided is a command line tool, which, given a class skeleton and a test suite,

will generate successive class implementations until all tests pass.

This work is heavily based off of the work in [Mohannadcse/AlloySpecRepair](https://github.com/Mohannadcse/AlloySpecRepair).

## Dependencies

* `ollama`

* `llama3:latest`

* `poetry`

### Serving Your Own LLM

In order to run the code as-is, you must self serve your own LLM locally.

Currently, this is hard-coded to llama3:latest.

If you have ollama installed, self serving this LLM locally is trivial:

```commandline

ollama serve

ollama run llama3:latest

```

## Running The Tool

For usage information, run:

```commandline

poetry run python -m main --help

```

# Roadmap

* Containerization

* Configurable OpenAI API key for ChatGPT usage