instancelib-onnx 0.1.3

A wrapper for ONNX models that adheres to the instancelib specification

instancelib-onnx

ONNX extension for instancelib.

Installation

You can install this package as follows:

pip install instancelib-onnx

Or by cloning this repo and issuing:

python setup.py

You will need at least Python 3.8 to use this library.

Usage

import instancelib as il
import ilonnx

# Specify the model location and the label translation 
model = ilonnx.build_data_model("example_models/data-model.onnx", 
                                {0: "Bedrijfsnieuws", 1: "Games", 2: "Smartphones"})

Then you can use the normal instancelib functionality to interact with the model.

# Load a dataset with instancelib
env = il.read_excel_dataset("datasets/testdataset.xlsx", ["fulltext"], ["label"])

# Assess the performance like any other instancelib model
performance = il.classifier_performance(model, env.dataset, env.labels)
performance.confusion_matrix