Neural Network Visualizer (Placeholder)
Real-time training insights with animated node graphs and metric overlays.
Interactive tooling to watch neural networks train with animated layers, metrics, and D3-driven visuals.


Quick Stats
Links
Neural Network Visualizer
Full case study refresh is coming soon—treat this as an early snapshot while I finish polishing copy and assets.
This project turns raw model data into an interactive playground where you can watch layers activate, weights adjust, and metrics evolve in real time.
Architecture
Data streams from TensorFlow.js into a worker thread, which keeps the visualization buttery smooth even when training larger models.
- Train model in TensorFlow.js
- Stream weight matrices over
postMessage - Render D3 force layout with WebGL-backed nodes

Feature Highlights
- Animated force-directed graphs with gradient weight edges.
- Epoch timeline scrubber with playback controls.
- Heatmaps for layer activations.
- WebGL instancing for fast node rendering.
const simulation = forceSimulation(nodes)
.force("link", forceLink(links).id((d) => d.id))
.force("charge", forceManyBody().strength(-120))
.force("center", forceCenter(width / 2, height / 2))


Observability shouldn't stop at metrics dashboards—seeing the structure evolve unlocks intuition.
Have thoughts?
Curious what others see or think
Feel free to reach out or leave feedback
Share FeedbackPrefer email? joshuatjhie@pm.me