Neuro Graph Memory

Graph-based Continual Learning Memory Engine

Architecture, Design & Development by Franz Ayestarán / Collaborative Pair Programming with GitHub Copilot in VS Code

Active Brain: aivillage.ngm

Export NGM to LLM Dataset

Export Raw Graph Memory

Export raw graph memory JSONL. Use the editor tools to convert into training formats.


Build Training Dataset v1

Generates three files: .raw.jsonl, .tasks.jsonl, and .manifest.json


Training Manifest Viewer

Inspect build metadata, task counts, and split counts from a generated .manifest.json file.


Import Training Dataset to NGM

Create a new .ngm brain from a dataset .txt file.

Text imports use heuristic extraction and are stitched through an anchor concept to reduce disconnected graph islands.

Lower values keep scores conservative; higher values increase contrast between imported experiences.

Generated Datasets

File Size (bytes) Modified Edit Download Delete
aivillage_raw-jsonl.jsonl 645611 2026-10-16T17:41:33 Edit Download
aivillage_training_v1.raw.jsonl 645611 2026-10-16T17:21:20 Edit Download
aivillage_training_v1.tasks.jsonl 1061334 2026-10-16T17:21:20 Edit Download

Preview Dataset Records

Inspect the first N records before training.

Confirm deletion

Enter the configured delete password to continue.