Home▸Machine Learning & Neural Networks▸Graph Attention Networks 2D: Attention Coefficient Visualizer

Graph Attention Networks 2D: Attention Coefficient Visualizer

Interactive 2D visualizer for Graph Attention Networks (GAT): watch a node's attention coefficients get computed live via LeakyReLU scoring and softmax normalization across its neighbors, with a live bar chart of every alpha weight, multi-head attention, and an entropy sanity-check gauge. Drag to pan and scroll to zoom the graph.

Machine Learning & Neural Networks2DAdvanced60 FPS📱 Mobile-adapted⇄ 3D version

Details

CategoryMachine Learning & Neural Networks

RenderingCanvas 2D

DifficultyAdvanced

Target FPS60

ControlsSelect the query node and graph topology, choose the number of attention heads and the active head, drag the LeakyReLU slope, softmax sharpness and layout-spread sliders, click any node in the 2D graph to select it directly, drag to pan and scroll to zoom the graph view, and toggle auto-cycle to sweep through every node automatically.

graph neural networksattention mechanismGATsoftmaxdeep learningmessage passing

Related simulations

Graph Attention Network: Attention Matrix ViewInteractive 2D Graph Attention Network: watch a GAT layer compute LeakyReLU attention scores…

Graph Attention Network: Node ClassificationInteractive 3D Graph Attention Network: watch a GAT layer compute LeakyReLU attention…

2D Graph Neural Network Message Passing2D simulator of message passing in a Graph Neural Network: a graph diagram laid out natively… ⊞ All 226 in Machine Learning & Neural Networks →

Graph Neural Networks that use a fixed sum, mean or max to combine neighbor information treat every neighbor as equally important — Graph Attention Networks (GAT) instead learn, per edge, exactly how much weight each neighbor deserves. This simulator computes real GAT attention coefficients on a small graph laid out in 2D: pick a query node and watch a live LeakyReLU-scored, softmax-normalized attention distribution light up its incoming edges and populate a sorted bar chart, switch between independently-learned attention heads, and drag the LeakyReLU slope and softmax sharpness controls to see the distribution sharpen or flatten in real time, backed by a sum-check and entropy gauge that confirm the softmax normalization on every recompute.

⚙ Under the hood

Interactive 2D visualizer for Graph Attention Networks: pick a query node and watch its real LeakyReLU-scored, softmax-normalized attention coefficients light up its neighbors on a pannable, zoomable node-link graph, read the same weights off a live sorted bar chart and an entropy gauge, switch between independently-learned attention heads, and see the distribution sharpen or flatten live.

graph neural networksattention mechanismGATsoftmaxdeep learningmessage passing

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install