[GitHub] [madlib] kaknikhil commented on issue #443: DL: Add training for multiple models

[GitBox <gi...@apache.org>]

kaknikhil commented on issue #443: DL: Add training for multiple models
URL: https://github.com/apache/madlib/pull/443#issuecomment-534320848
 
 
   Some more comments from testing 
   
   1. If the input table does not exist, we get this error
   ```
   ERROR:  spiexceptions.UndefinedTable: relation "cifar10" does not exist
   LINE 2:                 FROM cifar10
                                ^
   QUERY:   SELECT gp_segment_id, sum(dependent_var_shape[1]) AS images_per_seg
                   FROM cifar10
                   GROUP BY gp_segment_id
   
   CONTEXT:  Traceback (most recent call last):
     PL/Python function "madlib_keras_fit_multiple_model", line 21, in <module>
       fit_obj = madlib_keras_fit_multiple_model.FitMultipleModel(**globals())
     PL/Python function "madlib_keras_fit_multiple_model", line 42, in wrapper
     PL/Python function "madlib_keras_fit_multiple_model", line 87, in __init__
     PL/Python function "madlib_keras_fit_multiple_model", line 146, in get_image_count_per_seg_for_minibatched_data_from_db
   PL/Python function "madlib_keras_fit_multiple_model"
   ```
   We should instead error out with a meaningful error message just like in madlib_keras_fit.
   
   2. If the input table is not minibatched, then it fails with
   ```
   cifar10=# SELECT madlib.madlib_keras_fit_multiple_model('cifar10_train','cifar_multi_model','mst_table',6,0 );
   ERROR:  spiexceptions.UndefinedColumn: column "dependent_var_shape" does not exist
   LINE 1:  SELECT gp_segment_id, sum(dependent_var_shape[1]) AS images...
                                      ^
   QUERY:   SELECT gp_segment_id, sum(dependent_var_shape[1]) AS images_per_seg
                   FROM cifar10_train
                   GROUP BY gp_segment_id
   
   CONTEXT:  Traceback (most recent call last):
     PL/Python function "madlib_keras_fit_multiple_model", line 21, in <module>
       fit_obj = madlib_keras_fit_multiple_model.FitMultipleModel(**globals())
     PL/Python function "madlib_keras_fit_multiple_model", line 42, in wrapper
     PL/Python function "madlib_keras_fit_multiple_model", line 87, in __init__
     PL/Python function "madlib_keras_fit_multiple_model", line 146, in get_image_count_per_seg_for_minibatched_data_from_db
   PL/Python function "madlib_keras_fit_multiple_model"
   
   ```
   We should error out the same way we do in madlib_keras_fit i.e. `
   madlib_keras_fit error: Input table 'cifar10_train_summary' does not exist. Please ensure that the source table (cifar10_train) has been preprocessed by the image preprocessor.`

----------------------------------------------------------------
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