Downloads · 30 days
0
heterogat-v2
archi829/heterogat-v2
heterogat-v2 is a machine learning model from archi829. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as mit.
HeteroGAT v2 is a heterogeneous graph attention network designed to detect bottlenecks and anomalous patterns in system trace graphs.
Access
Public
Updated Apr 23, 2026
Repo size
260 KB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt260 KB · 99%
At a glance
- License
- mit
- Access
- Public
- Created
- Apr 23, 2026
- Updated
- Apr 23, 2026
- SHA
- bd6ec628
Datasets
Model card
From the Hugging Face model README
HeteroGAT v2 — Bottleneck Oracle Model
Overview
HeteroGAT v2 is a heterogeneous graph attention network designed to detect bottlenecks and anomalous patterns in system trace graphs.
It operates on graph-structured data derived from execution traces and learns node/edge relationships to identify suspicious or high-impact components.
Model Details
- Architecture: Heterogeneous Graph Attention Network (HeteroGAT)
- Framework: PyTorch + PyTorch Geometric
- Model File:
heterogat_v2.pt
Key Components
- Multi-head attention on heterogeneous node types
- Learned embeddings for different entity types
- Graph-level aggregation for prediction
Dataset
Trained on: 👉 https://huggingface.co/datasets/archi829/bottleneck-oracle-data
Input
- Preprocessed graph objects (
graphs_v2.pkl) - Nodes represent system entities (services, processes, etc.)
- Edges represent interactions/dependencies
Training Pipeline
- License
- mit
- Created
- Apr 23, 2026
- Updated
- Apr 23, 2026