Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
143 lines (104 loc) · 3.1 KB

CONTRIBUTING.md

File metadata and controls

143 lines (104 loc) · 3.1 KB

Contributing to vLLM-lite

Thank you for your interest in contributing to vLLM-lite! This guide will help you get started.

Prerequisites

Before building from source, ensure you have:

  • Rust 1.75+ - Install via rustup
  • CUDA 12.1+ - Optional, for GPU support
  • CMake 3.18+ - For building some dependencies
# Verify installation
rustc --version   # Should be 1.75 or higher
cargo --version

Quick Start

# Clone the repository
git clone https://github.com/pplmx/vllm-lite.git
cd vllm-lite

# Build
cargo build --workspace

# Run tests
cargo test --workspace

# Run the server
cargo run -p vllm-server

Development Workflow

  1. Fork & Clone

    • Fork the repository on GitHub
    • Clone your fork: git clone https://github.com/YOUR_USERNAME/vllm-lite.git

  2. Create a Branch

    git checkout -b feature/your-feature-name
# or
git checkout -b fix/bug-description

  3. Make Changes

    • Follow the coding standards below
    • Add tests for new features
    • Keep commits atomic and focused

  4. Test Your Changes

    # Format check
cargo fmt --all --check

# Lint
cargo clippy --workspace -- -D warnings

# Run tests
cargo test --workspace

  5. Submit a Pull Request

    • Push to your fork
    • Open a PR against main
    • Fill out the PR template

Coding Standards

  • Formatting: Run cargo fmt --all before committing
  • Linting: cargo clippy --workspace -- -D warnings must pass
  • Testing: Add tests for new features; all tests must pass
  • Documentation: Document public APIs with /// doc comments

Commit Message Format

<type>(<scope>): <subject>
Type Description
feat New feature
fix Bug fix
refactor Code restructuring
test Adding/updating tests
docs Documentation
chore Maintenance

Example:

feat(scheduler): add decode-priority batching

- Prioritize decode sequences over prefill
- Add max_num_batched_tokens limit
- Fix chunked prefill tracking

Project Structure

vllm-lite/
├── crates/
│   ├── traits/      # Interface definitions
│   ├── core/        # Engine, Scheduler, KV Cache
│   ├── model/       # Model implementations, kernels
│   ├── dist/        # Tensor Parallelism
│   └── server/      # HTTP API
├── tests/           # Integration tests
└── docs/            # Design documents

Testing

# Run all tests
cargo test --workspace

# Run specific crate
cargo test -p vllm-core

# Run with output visible
cargo test --workspace -- --nocapture

# Run only fast tests (skip slow/ignored tests)
just nextest

Getting Help

  • Issues: Open a GitHub issue for bugs or feature requests
  • Discussions: Use GitHub Discussions for questions
  • Documentation: See README.md and docs/ directory

License

By contributing, you agree that your contributions will be licensed under the MIT License.