ryanthompson591 opened a new issue, #21769: URL: https://github.com/apache/beam/issues/21769
### What needs to happen? Currently TFX, Scikit learn and Pytorch might get different types of input data. They then look at the type of input from the pcollection given at run time and decision what sort of path they will go on. Also these transforms use a large ever expanding list of typehints as the input type. The ideal way to resolve both these issues would be to allow the user (who should be aware of the input type) a way to pass that in to the model loader (which will return an inferer). Something like: model_loader = SklearnModelLoader(url='http://mymodel.com/model.pkl', input_type=pandas.DataFrame) model_loader = PytorchModelLoader(url='http://mymodel.com/model.pkl', input_type=dict) Issue Priority Priority: 2 Issue Component Component: sdk-py-core Subtask of issue https://github.com/apache/beam/issues/21435 ### Issue Priority Priority: 2 ### Issue Component Component: sdk-py-core -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
