[MADLIB-1371] DL: Exception during madlib_keras_fit when model_arch_id is passed as NULL

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: v1.17
• Component/s: Deep Learning
• Assignee: Unassigned
• Reporter: Nikhil Kak
• Created: 23/Jul/19 21:46
• Updated: 13/Aug/19 23:17
• Resolved: 06/Aug/19 20:54

Description

While calling madlib_keras_fit, if we pass in the model arch id as NULL, we get the following exception

SELECT madlib_keras_fit(
 'cifar_10_sample_batched',
 'keras_saved_out',
 'model_arch',
 NULL, --null model arch
 $$ optimizer=SGD(lr=0.01, decay=1e-6, nesterov=True), loss='categorical_crossentropy', metrics=['mae']$$::text,
 $$ batch_size=2, epochs=1, verbose=0 $$::text,
 3,
 NULL);


LINE 1: ...el_arch, model_weights FROM model_arch WHERE model_id = None
                                                                   ^
QUERY:  SELECT model_arch, model_weights FROM model_arch WHERE model_id = None
CONTEXT:  Traceback (most recent call last):
  PL/Python function "madlib_keras_fit", line 21, in <module>
    madlib_keras.fit(**globals())
  PL/Python function "madlib_keras_fit", line 42, in wrapper
  PL/Python function "madlib_keras_fit", line 95, in fit
PL/Python function "madlib_keras_fit"

 

 We should instead error out gracefully and print a more meaningful error message