[web] cannot run simple scikit-learn classifier in browser

[JulesFrancoise]

Hi,

I am trying to experiment with ORT web, however I get error messages whenever running inference.
I trained a simple Random Forest Classifier on IRIS following this example code: https://onnx.ai/sklearn-onnx/auto_examples/plot_convert_model.html#sphx-glr-auto-examples-plot-convert-model-py

I am then trying to run predictions in the browser with a modification of this example: https://github.com/microsoft/onnxruntime-inference-examples/tree/main/js/quick-start_onnxruntime-web-script-tag
Where I use as input:

// prepare inputs. a tensor need its corresponding TypedArray as data
const dataA = Float32Array.from([1, 2, 3, 4]);
const tensorA = new ort.Tensor('float32', dataA, [1, 4]);

// prepare feeds. use model input names as keys.
const feeds = { float_input: tensorA };

However, with any model (except LinearRegression) I get cryptic error messages from session.run() like 8978344
What am I missing? Can anyone help me?

Thanks in advance.

[JulesFrancoise]

I managed to solve my problem...

First, I tried to run the pipeline with onnxruntime-node, where I got a readable error message:
failed to inference ONNX model: Error: Non tensor type is temporarily not supported..

After searching around issues and documentation, I understood that the zipmap option needs to be deactivated when converting Scikit-Learn classifiers with sklearn-onnx, otherwise none of the classifiers will work with onnxruntime in JavaScript.