Standalone Example

Standalone Example (brain_standalone.jl)

Hardware validation script that exercises production APIs outside the test harness.

Purpose

• Confirm CUDA device visibility.
• Construct SparseBrain and EnsembleBrain with custom n_in / n_out.
• Run multi-tick loops with inhibition and reflex gating.
• Print diagnostics and readouts.

How to Run

# from repo root, with GPU
julia --project examples/brain_standalone.jl

Requires CUDA.functional() — will fail on CPU-only VMs when constructing brains.

Sequence

1. Print CUDA.name(CUDA.device()).
2. SparseBrain(20.0f0; n_in=8, n_out=4, name="standalone").
3. Zero input u = CUDA.zeros(Float32, 8).
4. 100 step!(brain, u; inhibition=0.2f0) iterations.
5. Print diagnostics and get_output.
6. EnsembleBrain(n_in=8, n_out=4).
7. 50 ensemble_step!(ensemble, u; inhibition=0.1f0, reflex_signal=0.15f0) — reflex active (0.15 > 0.1).
8. Print ensemble_diagnostics and get_ensemble_output.

Dependencies

using LiquidCortex
using CUDA

Uses the project environment (--project).

Interpreting Output

• Non-zero rate= percentages after steps indicate spiking under noise + threshold.
• With all-zero input, activity comes mainly from OU noise and recurrent dynamics.
• Ensemble lines show per-lobe τ, tick, rate, and W_out norm.

Related

• Getting Started
• Test Suite
• Neural Execution API

---

Last updated: July 28, 2026 Updated by: Grok Build: Grok 4.5 Package tip reference: 4e2698c (main)