Skip to content

Latest commit

 

History

History
48 lines (31 loc) · 2.2 KB

README.md

File metadata and controls

48 lines (31 loc) · 2.2 KB

CodeContrast

CodeContrast is a novel generative model that leverages contrastive learning to map programming problems, test cases, and code solutions into a shared feature space. By minimizing the distance between matching components and maximizing the distance between non-matching components, CodeContrast captures the intricate relationships between these elements, enabling the generation of coherent and aligned programming exercises.

Features

  • Coherent Exercise Generation: CodeContrast generates programming exercises where the problem descriptions, test cases, and code solutions are semantically aligned and coherent.
  • Diverse Exercise Generation: The model can generate diverse programming exercises across various problem domains and programming concepts.
  • High Code Correctness: The generated code solutions exhibit a high degree of correctness, passing a significant portion of test cases.
  • Comprehensive Test Cases: The generated test cases provide comprehensive coverage, ensuring thorough evaluation of the generated solutions.
  • Pedagogical Value: The generated exercises are tailored for introductory programming courses and have been validated for their pedagogical value by experts and student studies.

Installation

  1. Clone the repository:

    git clone https://github.com/nicolastorresr/CodeContrast.git

  2. Install the required dependencies:

    cd CodeContrast
pip install -r requirements.txt

Usage

  1. Prepare your data: Ensure that you have a dataset of programming problems, test cases, and code solutions in the appropriate format.

  2. Configure the model: Update the configuration file config.py with your desired settings, such as the paths to pretrained models, hidden sizes, and other hyperparameters.

  3. Train the model:

    python train.py

    This will start the training process for the CodeContrast model. You can monitor the training progress and loss values printed to the console.

  4. Generate programming exercises:

    python generate.py

    This script will use the trained CodeContrast model to generate programming exercises, including problem descriptions, test cases, and code solutions.