This model is intended to work with the Python 3 ONNX Drop-In Environment.
In addition to the main model artifact, this model also has an artifact for the preprocessing pipeline. Because of these extra artifacts and preprocessing, custom.py uses the load_model and transform hooks at predict time.
The supplied .onnx file is the ONNX export of a PyTorch model trained on skyserver_sql2_27_2018_6_51_39_pm.csv
with class as the target (multiclass).
Paths are relative to datarobot-user-models root:
drum score --code-dir model_templates/python3_onnx_multiclass --input tests/testdata/skyserver_sql2_27_2018_6_51_39_pm.csv --target-type multiclass --class-labels-file model_templates/python3_onnx_multiclass/class_labels.txt