You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@madlib.apache.org by GitBox <gi...@apache.org> on 2019/05/22 21:30:30 UTC

[GitHub] [madlib] kaknikhil opened a new pull request #396: Dl transfer learning

kaknikhil opened a new pull request #396: Dl transfer learning
URL: https://github.com/apache/madlib/pull/396
 
 
    DL: Add model_weights, name and desc to the model arch table.
   
   JIRA: MADLIB-1348
   
   This commit add three optional params to the model arch interface to
   enable transfer learning
   1. model weights in bytea format
   2. name
   3. description
   
   Additionally
   1. Use plpy prepare to format the model weight as bytea.
   2. Also had to modify the deserialize code to accept a bytea string
   instead of a double precision[]
   3. Modified madpack code so that install check user can create python
   UDFs
   4. Add dev check test to test for transfer learning by creating a UDF
   that calls the load_arch_table with some pre defined weights so that the
   first iteration of fit always returns the same loss and metric.
   5. Rename Format class with a more meaningful name.
   

----------------------------------------------------------------
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.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services