Developer Console

Inference Playground

Load an .onnx model, set the input tensor, and run it with the Rust WASM runtime in this tab or with the native C++ engine on the server. The WASM runtime is a separate small interpreter (MatMul, Add, Relu, Sigmoid, Softmax); CNNs and other ops need the native engine.

Inference Console

Backend

Drop .onnx model here

Loaded: —

Input shape

[]

Expecting 1 float32 values.

Input values

✗ Got 0 values — shape [1] needs 1

The fraud model expects 7 z-score-normalised features (zeros = the training mean). The Fraud Demo page does this normalisation from raw amounts for you.

Model Graph Inspector

Loading model…