Model testing before uploading

When you are creating your own models, it is essential to check the model (and its accuracy on some test-set) before deploying the model to the Scailable platform.
On a local machine or in the cloud, ONNX models are easily tested using the ONNX runtime.
The sclblonnx package provides and easy wrapper around the ONNX runtime for usage on your local machine directly from python. The following code snippet provides the gist of it:
import sclblonnx as so
// code to construct graph g with outuput node output and input example e.
result = so.run(g,
inputs=e,
outputs=["output"]
)
print(result)
Always check whether your input provides the desired output. Examples of usage of the run() function can be found here.
Note that the conversion of ONNX, when uploaded to the Scailable platform, to SPMF is one-to-one. Thus, the model output (and its performance in terms of accuracy) on the edge device is the exact same as you will find locally using the sclblonnx package.