Star ⭐ this repository if you find it useful!
This repository contains the first implementation of AlphaCodium in the LangChain framework.
AlphaCodium is an advanced code generation tool that enhances the performance of Large Language Models (LLMs) on complex coding tasks. It introduces a multi-stage, iterative process specifically tailored for code generation, which has shown significant improvements in accuracy. For instance, AlphaCodium increased GPT-4's accuracy from 19% to 44% on the pass@5 metric, outperforming existing methods like DeepMind's AlphaCode.
Currently, the AI test case generation and iterative AI test case generation implementations are pending and will be completed soon. The rest of the paper's proposed workflow has been implemented using the original prompts provided in the paper.
-
Clone the repository:
git clone https://github.com/hamidriasat/AlphaCodium.git cd AlphaCodium -
Install the required dependencies:
pip install -r requirements.txt
-
Place your OpenAI API key in the .env file.
-
Execute the alphacodium_ai_code_assistant.ipynb file to test AlphaCodium.
Note: The RetryOutputParser is not implemented yet. Errors may occur when the LLM fails to generate the required complex output structure, resulting in OutputParser errors. In such cases, rerun the cell until the LLM produces the output in the correct format.
- Complete README.md
- Restructure overall code
- Add retry output parser
- Add AI test generation + Iterative AI test execution steps
We appreciate any feedback so reporting problems, and asking questions are welcomed here.
This implementation is a work in progress and not an official version of AlphaCodium. It aims to explore the integration of AlphaCodium within the LangChain framework, leveraging its robust architecture to enhance code generation capabilities.