This example shows:
- How to create an InferenceSession using
Node
- Load a simple ONNX model which has one Operator: Add
- Create two random Tensors of given shape
- Run the inference using the inputs
- Get output Tensor back
- Access raw data in the Tensor
- Match the results against the exptected values
-
Download model file
add.onnx
from examples/models and put it in the current folder. -
Inside the folder, run
npm install
. -
run
node index.js
-
index.js
The main .js file that holds all
ONNX.js
logic to load and execute the model. -
package.json
A package file that specifies
ONNX.js
dependency. -
add.onnx
A simple ONNX model file that contains only one
Add
operator.