Project Synapse is currently maintained as a research codebase. Contributions are welcome, especially in the following areas:
- Reproducibility and experiment cleanup
- Resource-aware conditioning with real telemetry
- Adapter and LoRA baselines
- Evaluation scripts and benchmark automation
- Test coverage for additional experiment paths
conda env create -f 02_environment.yml
conda activate synapseOr:
python -m venv .venv
source .venv/bin/activate
pip install -r 03_requirements.txtpython -m unittest discover -s tests -v
python scripts/train.py --config configs/public_toy_exp.yaml
python scripts/train.py --config configs/public_digits_exp.yaml
python scripts/train.py --config configs/public_digits_resource_exp.yaml- Keep changes focused and well-scoped.
- Prefer adding or updating tests when behavior changes.
- Document limitations honestly.
- Preserve the repository's research-oriented tone and modular structure.