Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
65 lines (48 loc) · 1.75 KB

CONTRIBUTING.md

File metadata and controls

65 lines (48 loc) · 1.75 KB

Contributing

Thanks for your interest in improving gemma-realtime. Here's how to help.

Areas Where Help Is Needed

High Impact:

  • More data extractors (Telegram, Discord, Signal, WhatsApp native API)
  • CoreML/ANE optimization for the draft model (currently falls back to MLX)
  • Windows/Linux support (currently macOS-focused due to Apple Silicon)
  • Voice pipeline integration (STT + inference + TTS end-to-end)

Medium Impact:

  • Better auto-detection of PLE-safe model variants
  • Training data quality metrics and filtering
  • Multi-GPU support for M-series Ultra chips
  • Benchmark result archiving and regression tracking

Documentation:

  • Guides for additional messaging platforms
  • Video walkthrough of the full pipeline
  • Troubleshooting guide for common issues

Development Setup

git clone https://github.com/sethdford/gemma-realtime.git
cd gemma-realtime
pip install mlx mlx-lm

Code Style

  • Python: standard library preferred, minimal dependencies
  • Shell: bash 3.2 compatible (macOS default), no associative arrays
  • Scripts should be self-contained and runnable independently

Testing Changes

# Verify all scripts parse correctly
python3 -m py_compile scripts/*.py
bash -n scripts/*.sh

# Run the benchmark
python3 scripts/voice-bench.py --rounds 5

Pull Requests

  1. Fork the repo
  2. Create a feature branch (git checkout -b feat/telegram-extractor)
  3. Make your changes
  4. Test on Apple Silicon if possible
  5. Open a PR with a clear description

Privacy

Never commit:

  • Training data (JSONL files with personal messages)
  • Model adapters (.safetensors files)
  • Benchmark results that contain message previews
  • API keys or credentials

The .gitignore already excludes these, but please verify before committing.