gnn-environment-setup
GNN environment setup and dependency management. Use when configuring the development environment, installing dependencies, managing virtual environments, or troubleshooting dependency issues for the GNN pipeline.
安装与使用
请根据来源仓库和你的 AI 工具说明安装此技能。先阅读技能文件和依赖要求,再添加到工作流。
查看来源 ↗使用指南技能说明
GNN Environment Setup (Step 1)
Purpose
Manages environment configuration, virtual environment creation, and dependency installation for the GNN pipeline. Supports both core and optional dependency groups via uv or pip.
Key Commands
# Run setup step
python src/1_setup.py --target-dir input/gnn_files --output-dir output --verbose
# As part of pipeline
python src/main.py --only-steps 1 --verbose
# Using uv (recommended)
uv sync # Core dependencies
uv sync --extra dev # Development tools
uv sync --all-extras # Everything
Optional Dependency Groups
These groups match [project.optional-dependencies] in pyproject.toml:
| Group | Key Packages | Purpose |
|---|---|---|
active-inference |
pymdp, jax, jaxlib, flax, optax | Active Inference simulation |
visualization |
plotly, altair, seaborn, bokeh, holoviews | Visualization |
llm |
openai, anthropic, cohere, ollama | LLM integration |
audio |
librosa, soundfile, pedalboard, pydub | Audio generation |
gui |
gradio, streamlit | GUI interface |
ml-ai |
torch, transformers, datasets | Machine learning |
graphs |
igraph, graphviz, discopy | Graph libraries |
dev |
pytest-*, mypy, ruff | Development tools |
database |
sqlalchemy, alembic | Database integration |
research |
jupyterlab, sympy, numba, cython | Research tools |
scaling |
dask, distributed, joblib | Parallel processing |
all |
Everything above | Full installation |
Environment Requirements
- Python: 3.11+ (recommended 3.12.x or 3.13.x)
- Package manager: uv (recommended) or pip
- Core deps: numpy, networkx, pyyaml, jsonschema
Troubleshooting
| Issue | Solution |
|---|---|
uv: command not found |
curl -LsSf https://astral.sh/uv/install.sh | sh |
| PyMDP import fails | uv sync --extra active-inference |
| matplotlib missing | uv sync --extra visualization |
| CUDA not detected | Check torch.cuda.is_available() after uv sync --extra ml-ai |
MCP Tools
This module registers tools with the GNN MCP server (see mcp.py):
-
ensure_directory_exists -
find_project_gnn_files -
get_standard_output_paths
References
- AGENTS.md — Module documentation
- README.md — Usage guide
- SPEC.md — Module specification
- ../../SETUP_GUIDE.md — Full setup guide
- ../../pyproject.toml — Dependency definitions
Documentation
使用示例
查看来源仓库中的使用示例与配置说明。
查看来源 ↗来源历史
- 收录时间
- 2026-03-11
- 索引更新时间
- 2026-03-11
- Branch
- main
收录与同步日期属于本站记录。源码修改历史请查看来源仓库。
GitHub ↗chat Comments (0)
Sign in to join the discussion and leave a comment.
No comments yet. Be the first to share your thoughts!