python-gfm 0.1.8

An AI copilot for graph data and models (Under active development).

pygfm

PyPI version Python License

A unified Python toolkit for Graph Foundation Models (GFM) research, integrating 19 baseline methods under a single package with shared utilities, standardized interfaces, and reproducible experiment pipelines.

Developed by Beihang University, School of Computer Science and Engineering, ACT Lab, MAGIC GROUP.

Installation

pip install python-gfm

Install with optional dependencies:

pip install python-gfm[torch]   # PyTorch + PyG
pip install python-gfm[dev]     # pytest + ruff

For development (full checkout with experiment scripts):

git clone <repo-url> && cd pythongfm
pip install -e ".[torch,dev]"

Quick Start

import pygfm

print(pygfm.__version__)

Package Structure

pygfm/
├── baseline_models/   # 19 GFM baseline implementations
├── public/            # Shared utilities, losses, backbone encoders
│   ├── backbone_models/
│   ├── utils/
│   ├── cli/
│   └── model_bases.py
├── private/           # Internal modules (core encoders, data generation)
└── cli/               # Console entrypoints

Supported Baselines

Category	Baselines
Prompt-based GFM	MDGPT, SAMGPT, MDGFM, GraphPrompt, HGPrompt, MultiGPrompt, GCoT
Structure-aware	SA2GFM, Bridge, GraphKeeper, GraphMore, Graver, BIM-GFM
LLM-integrated	GraphGPT, GraphText, LLaGA, OneForAll
Retrieval-augmented	RAG-GFM
Classic	Classic GNN

Running Experiments

Each baseline has its own experiment scripts under scripts/<baseline>/. Run from the repository root:

# Pre-training
python scripts/mdgpt/pretrain.py

# Downstream fine-tuning
python scripts/sa2gfm/downstream.py

# Full pipeline with config
python scripts/gcot/pretrain.py
python scripts/gcot/finetune.py
python scripts/gcot/finetune_graph.py

Console Commands

After installation, the following CLI commands are available:

Command	Description
gfm-sa2gfm-pretrain	SA2GFM contrastive pre-training (-c YAML)
gfm-sa2gfm-downstream	SA2GFM MoE downstream fine-tuning

Baseline Documentation

Detailed instructions for each baseline:

Baseline	Docs
MDGPT	[scripts/mdgpt/README.md](scripts/mdgpt/README.md)
SA2GFM	[scripts/sa2gfm/README.md](scripts/sa2gfm/README.md)
MultiGPrompt	[scripts/multigprompt/README.md](scripts/multigprompt/README.md)
LLaGA	[scripts/llaga/README.md](scripts/llaga/README.md)
GraphText	[scripts/graphtext/README.md](scripts/graphtext/README.md)
GraphGPT	[scripts/graphgpt/README.md](scripts/graphgpt/README.md)
OneForAll	[scripts/oneforall/README.md](scripts/oneforall/README.md)
MDGFM	[scripts/mdgfm/README.md](scripts/mdgfm/README.md)
SAMGPT	[scripts/samgpt/README.md](scripts/samgpt/README.md)
GCoT	[scripts/gcot/README.md](scripts/gcot/README.md)
HGPrompt	[scripts/hgprompt/README.md](scripts/hgprompt/README.md)
GraphPrompt	[scripts/graphprompt/README.md](scripts/graphprompt/README.md)
Graver	[scripts/graver/README.md](scripts/graver/README.md)
GraphMore	[scripts/graphmore/README.md](scripts/graphmore/README.md)
Bridge	[scripts/bridge/README.md](scripts/bridge/README.md)
GraphKeeper	[scripts/graphkeeper/README.md](scripts/graphkeeper/README.md)
RAG-GFM	[scripts/rag_gfm/README.md](scripts/rag_gfm/README.md)

Configuration

Experiment configs are YAML files located at scripts/<baseline>/configs/. Pass them via -c flag or as positional arguments depending on the baseline.

Important: Do not commit API keys. For baselines that require LLM API access (e.g., GraphText), copy the example config and fill in your keys locally:

cp scripts/graphtext/config/user/env.yaml.example scripts/graphtext/config/user/env.yaml

License

This project is licensed under the Apache License 2.0.

Team

MAGIC GROUP -- Beihang University, School of Computer Science and Engineering, ACT Lab.