ERROR Message when using and importing Machine Learning model
0
When trying to work with the online mendix tutorial (A Guide to ML-Enhanced Applications Powered by Mendix and Amazon SageMaker), which uses the TF-IDF Vectorizer scikit-learn algorithm we can’t get the model imported into Mendix.We have transformed it into ONNX just like the tutorial did, but we get the following message: “ML model file ‘C:\Users\...\Downloads\spam_nb.onnx’ wasn’t imported due to a technical issue.” We really have no idea why we get this.. ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Different error: Currently I’m doing a project that involves machine learning. For this we want to use a Keras machine learning model. The problem is the error we encounter after we try to send the input through the application. “An error occurred, please contact your system administrator.” With the following error details, that vary depending on the input (1 word, 2 words, 1 word but a different error) We're using the following microflow to trigger the machine learning model once the button "save" is pressed After importing the model the mapping is as followed: In the input we've tried to change the mapping so that the input is "String" instead of a flaot with a "toString" function but this causes the model to give an error when importing it. Just like the one in the first picture. If there is anything that you might know about these errors please let us know. Thanks!
asked
Thijs Onderwater
1 answers
0
Hi Thijs Onderwater,
Based on microflow I see that you have not used pre-processing action which should be a java Action which converts your input into ml model input.
Based on Ml model the input is embedding_1_input which is a array of 1*20 size But in mendix we can only pass it as string(Embedding_1_input). so, we need to use a java action which can convert your input in a format of array and change this array as string and make the java action return the object Ml_input_Entity which is this converted string.
Hope this helps, if you need further assistance feel free to reply.
answered
Ganesh Pavan Munduri