Home / cuhk-aim-group / neurodiscovery · GitHub

cnn3d SKILL.md skillA

cnn3d from cuhk-aim-group/neurodiscovery is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned

mdr add cuhk-aim-group/neurodiscovery/cnn3d@git:20261023.f74c2bd

Pins git:20261023.f74c2bd in mdr.lock. Pin the fingerprint instead.

Audit grade: A, 17 of 17 checks passed
Kind: SKILL.md skill
Repository: cuhk-aim-group/neurodiscovery
Path: skills/cnn3d/SKILL.md
Versions recorded: 2
Latest version: git:20261023.f74c2bd, committed Oct 23, 2026
First version recorded: Aug 25, 2026
Size: 2,972 bytes
Repository stars: 93
License: MIT

Use this model skill whenever the user wants a compact residual 3D convolutional neural network for voxel-level classification or regression from structural MRI, functional MRI summaries, statistical maps, or other aligned volumetric neuroimaging data. Triggers include 'CNN3D', '3D CNN', 'voxel model', 'volumetric MRI', 'whole-brain volume classification', 'sMRI deep learning', and 'voxel regression'.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# CNN3D Skill

## Overview

`cnn3d` is NeuroRuntime's canonical compact residual 3D CNN. It owns one model,
one NPZ input contract, and one checkpoint format. NeuroSTORM remains a
separate model skill with its own external repository and runtime.

| Model | Input | Tasks |
|---|---|---|
| `VoxelCNN3D` | whole-volume tensor | classification, regression |

---

## Installation

```bash
pip install numpy torch scikit-learn pandas
```

Verify:

```bash
python -c "from models.cnn3d import VoxelCNN3D; print('CNN3D OK')"
```

---

## Workflows

### 1. Prepare a volume NPZ

```text
X:          float array [subjects, channels, depth, height, width]
y:          array [subjects]
subject_id: string array [subjects] (optional)
```

All subjects must use the same orientation, voxel size, grid, crop, and
intensity-normalization protocol.

### 2. Classification

```bash
python skills/cnn3d/scripts/train_reference.py \
  --input volumes.npz \
  --task classification \
  --base-channels 16 \
  --dropout 0.1 \
  --epochs 50 \
  --batch-size 4 \
  --folds 5 \
  --device cuda \
  --output-dir run_models_output/cnn3d
```

### 3. Regression

```bash
python skills/cnn3d/scripts/train_reference.py \
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint

mdr add cuhk-aim-group/neurodiscovery/cnn3d@sha256:65b916977aaab7ff

The line at the top of this page pins the label git:20261023.f74c2bd, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

[![mdr](https://markdownregistry.com/badge/art_qbhhoet5oo4odhzl.svg)](https://markdownregistry.com/a/art_qbhhoet5oo4odhzl)

4 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 2 versions of this file, between Aug 25, 2026 and Oct 23, 2026, about one every 59 days.

version	committed	commit	size	audit
git:20261023.f74c2bd latest	2026-10-23	f74c2bd	2,972 B	A	view · diff
git:20260825.cbd02ff	2026-08-25	cbd02ff	2,969 B	A	view

Audit of the latest version

A 17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.

pass: Frontmatter block present
pass: Frontmatter declares a name
pass: Frontmatter declares a description
pass: Size between 200 bytes and 200 KB (2972 bytes)
pass: No zero-width or bidi control characters
pass: No instruction hidden inside an HTML comment
pass: No link to an exfiltration or paste host
pass: No credential-shaped string
pass: No instruction to send local credentials anywhere
pass: No text hidden with inline styles
pass: No prompt-injection phrasing
pass: No curl or wget piped into a shell
pass: No recursive delete of root, home or parent
pass: No instruction to read or print local credentials
pass: No base64 blob over 200 characters
pass: No link to a raw IP address
pass: No script tag

Source

GitHub

cuhk-aim-group/neurodiscovery · 93 stars · license MIT · pushed 2026-10-23 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_qbhhoet5oo4odhzl
GET https://markdownregistry.com/api/v1/resolve?ref=cuhk-aim-group/neurodiscovery/cnn3d
GET https://markdownregistry.com/api/v1/blob/65b916977aaab7ffd169c0282548d38595311452f7894a47010c01aab584d29b

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.