Interactive Neuromorphic Computing & Spiking Neural Network Laboratory
The first comprehensive in-browser spiking neural network laboratory. Train neurons that communicate with spikes, not gradients. Explore the computational paradigm behind Intel's Loihi 3, IBM's NorthPole, and BrainChip's Akida 2.0.
In January 2026, Intel's Loihi 3 went mainstream — 8 million digital neurons, 64 billion synapses, 1000x more energy-efficient than GPUs. Neuromorphic computing is the biggest paradigm shift in computing since GPUs. Yet most people have zero intuition for how spiking neural networks work.
SpikeForge changes that with 7 deeply interactive modules.
-
LIF Neuron Playground — Leaky Integrate-and-Fire neuron with real-time membrane potential trace, phase portrait, and spike raster. Compare LIF, Izhikevich, and Adaptive Exponential models.
-
STDP Learning Visualizer — Spike-Timing Dependent Plasticity. Interactive pre/post spike timing, real-time STDP window curve, synaptic weight evolution.
-
Spike Train Encoder/Decoder — Rate coding vs temporal coding vs population coding. Encode analog signals into spike trains and decode them back.
-
Network Dynamics & Synchrony — Build networks of 50-500 spiking neurons. Observe emergent oscillations, raster plots, power spectra with neural frequency bands.
-
SNN Pattern Recognition — Train an SNN to recognize spatiotemporal patterns using only STDP — no backpropagation. Weight matrix heatmap and training curves.
-
Liquid State Machine — Reservoir computing with spiking neurons. Temporal XOR, parity check, and fading memory tasks with PCA state visualization.
-
Neuromorphic Energy Dashboard — Compare Loihi 3, NorthPole, Akida 2.0 vs H100 GPU. Event-driven vs clock-driven animation. CO2 savings calculator.
- Single HTML file, zero dependencies
- Canvas API for real-time neural visualization
- Runs entirely in the browser — no server needed
- ~93 KB total
MIT