[ https://issues.apache.org/jira/browse/MADLIB-1421?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17069042#comment-17069042 ]
Frank McQuillan commented on MADLIB-1421: ----------------------------------------- Could you please give an example of the workflow that you envision? NPZ is a compressed format for storing Numpy arrays https://docs.scipy.org/doc/numpy/reference/routines.io.html#npz-files and not (to my knowledge) for storying model descriptions in Keras or TensorFlow. Keras model description are typically stored in JSON and YAML https://keras.io/getting-started/faq/#savingloading-only-a-models-architecture so we decided to support JSON. Please provide a bit more information on what you would like here. Thanks > Support npz format > ------------------ > > Key: MADLIB-1421 > URL: https://issues.apache.org/jira/browse/MADLIB-1421 > Project: Apache MADlib > Issue Type: New Feature > Reporter: Hubert Zhang > Priority: Major > > Currently, we only support pass json format model. > Better to support npz format in future. -- This message was sent by Atlassian Jira (v8.3.4#803005)