CNTK: Python vs. C # API for Model Consumption

Question

We have prepared a model using CNTK. We are creating a service that will download this model and respond to requests for classification of offers. What is the best API to use regarding performance? We would rather build a C # service, as in https://github.com/Microsoft/CNTK/tree/master/Examples/Evaluation/CSEvalClient , but as an alternative, we plan to create a Python service that will load the model into python.

Do you have any recommendations regarding this or that approach? (regarding which API is faster, actively supported, or other parameters that you can think of). The next step would be to create an experiment that measures the performance of both API calls, but wondered if there was some preliminary knowledge here that could help us decide.

thank

Answer 1

Both APIs are well designed / maintained. For textual data, I would go with the C # API.

In C #, the main focus is quick and easy evaluation, and for text loading, the data is simple.

API Python / , . , - Python, #.

Answer 2

API # Eval CNTKLibrary ( -, , ). API API ++ Python .

API , , , .

turorial , Eval API ASP.Net. - EvalDLL, # API. # API.