训练与评测机器学习与科研 AIK-Dense-AI/claude-scientific-skills训练与评测
TO

Torch Geometric

维护者 K-Dense Inc. · 最近更新 2026年4月1日

Torch Geometric:PyTorch Geometric是一个库 built on PyTorch ,用于 developing 、 training Graph Neural Networks (GNNs)。 Apply this skill ,用于 深度学习 on graphs 、 irregular structures,,涵盖 mini-batch processing,multi-GPU training,、 geometric 深度学习 applications。

Claude CodeOpenClawNanoClaw训练编排评测比较torch-geometricmachine-learningpackagemachine learning & deep learning

原始来源

K-Dense-AI/claude-scientific-skills

https://github.com/K-Dense-AI/claude-scientific-skills/tree/main/scientific-skills/torch-geometric

维护者
K-Dense Inc.
许可
MIT license
最近更新
2026年4月1日

技能摘要

来自 SKILL.md 的关键信息

2 min

核心说明

  • PyG是一个standard 库 ,用于 Graph Neural Networks built on PyTorch. It provides data structures ,用于 graphs,60+ GNN layer implementations,scalable mini-batch training,、 support ,用于 heterogeneous graphs。
  • Install:uv add torch_geometric (或 uv pip install torch_geometric;requires PyTorch). Optional:pyg-lib,torch-scatter,torch-sparse,torch-cluster ,用于 accelerated ops。
  • edge_index = edge_pairs.t().contiguous()。
  • 用于 undirected graphs,include both directions:edge (0,1) needs both [0,1] 、 [1,0] in edge_index。
  • 用于 heterogeneous graphs,use HeteroData — see Heterogeneous Graphs section below。

原始文档

SKILL.md 摘录

Graph Data: Data and HeteroData

A graph lives in a Data object. The key attributes:

edge_index format is critical: it's a [2, num_edges] tensor where edge_index[0] = source nodes, edge_index[1] = target nodes. It is NOT a list of tuples. If you have edge pairs as rows, transpose and call .contiguous():


## Datasets

PyG bundles many standard datasets that auto-download and preprocess:

```python
from torch_geometric.datasets import Planetoid, TUDataset

## Single-graph node classification (Cora, Citeseer, Pubmed)

dataset = Planetoid(root='./data', name='Cora')
data = dataset[0]  # single graph with train/val/test masks

适用场景

  • Use Torch Geometric in 科研工作流 aligned ,支持 this subject area。
  • Follow upstream documentation ,用于 full working procedure。
  • **Graph-based 机器学习**:Node 分类,graph 分类,link prediction。
  • **Molecular property prediction**:Drug discovery,chemical property prediction。

不适用场景

  • Do not rely on this catalog entry alone ,用于 installation 或 maintenance details。

相关技能

相关技能

返回目录
CI
训练与评测机器学习与科研 AI

cirq

cirq:Cirq is Google Quantum AI's open-source 框架 ,用于 designing,simulating,、 running quantum circuits on quantum computers…

Claude Code训练编排
K-Dense-AI/claude-scientific-skills查看
GT
训练与评测机器学习与科研 AI

gtars

gtars:Gtars是一个high-performance Rust 工具包 ,用于 manipulating,analy。

Claude CodeOpenClaw训练编排
K-Dense-AI/claude-scientific-skills查看
PE
训练与评测机器学习与科研 AI

PennyLane

PennyLane是一个quantum computing 库 that enables training quantum computers like neural networks。 It provides automatic diff…

Claude Code训练编排
K-Dense-AI/claude-scientific-skills查看
PU
训练与评测机器学习与科研 AI

PufferLib

PufferLib是一个high-performance reinforcement learning 库 designed ,用于 fast parallel environment 模拟 、 training。 It achieves…

Claude Code训练编排
K-Dense-AI/claude-scientific-skills查看