[ 
https://issues.apache.org/jira/browse/MADLIB-1467?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Frank McQuillan closed MADLIB-1467.
-----------------------------------
    Resolution: Fixed

https://github.com/apache/madlib/pull/546

> madlib_keras_fit_multiple_model goes down with an IndexError: tuple index out 
> of range
> --------------------------------------------------------------------------------------
>
>                 Key: MADLIB-1467
>                 URL: https://issues.apache.org/jira/browse/MADLIB-1467
>             Project: Apache MADlib
>          Issue Type: Bug
>          Components: Deep Learning
>            Reporter: Pavan Kumar Nagula
>            Priority: Major
>             Fix For: v1.18.0
>
>
> SELECT 
> madlib.madlib_keras_fit_multiple_model('ml_ntwrk_training_final_dl_packed',  
> -- source_table
>                        'ml_ntwrk_multi_model',   -- model_output_table
>                        'mst_table',      -- model_selection_table
>                        10,           -- num_iterations
>                        TRUE,          -- use gpus
>                        'ml_ntwrk_validation_final_dl_packed',
>                        3,
>                        FALSE
>                        );
> ERROR: spiexceptions.InternalError: IndexError: tuple index out of range 
> (plpy_elog.c:121) (seg3 10.138.15.194:40003 pid=15416) (plpy_elog.c:121)
> CONTEXT: Traceback (most recent call last):
>  PL/Python function "madlib_keras_fit_multiple_model", line 25, in <module>
>   fit_obj.fit_multiple_model()
>  PL/Python function "madlib_keras_fit_multiple_model", line 280, in 
> fit_multiple_model
>  PL/Python function "madlib_keras_fit_multiple_model", line 319, in 
> train_multiple_model
>  PL/Python function "madlib_keras_fit_multiple_model", line 996, in 
> run_training
>  
>  
>  
>  
>  
> configauration in mst_table 
>  
> config  in mst_table  SELECT * FROM mst_table ORDER BY mst_key;
>  
> [4:39|https://pivotal.slack.com/archives/D01MS1EBQ4T/p1613032760009700]
> mst_key model_id compile_params fit_params
> 1 2 
> loss='categorical_crossentropy',optimizer='Adam(lr=0.001)',metrics=['accuracy']
> batch_size=128,epochs=3,
> class_weight={0: 96.67510706743768,
> 1: 7099.175362318841,
> 2: 26621.907608695652,
> 3: 4018.4011484823627,
> 4: 17747.9384057971,
> 5: 17.063958085855717,
> 6: 10141.679089026915,
> 7: 23663.917874396135,
> 8: 30425.037267080745,
> 9: 0.19866780178818547,
> 10: 91.95823008185026,
> 11: 70991.7536231884,
> 12: 53243.815217391304,
> 13: 806.7244729907774,
> 14: 20.452824437680324,
> 15: 21297.52608695652,
> 16: 13.401413344422679,
> 17: 0.07584896996158862,
> 18: 106487.63043478261,
> 19: 217.54367810987253,
> 20: 208.79927536231884,
> 21: 10648.76304347826,
> 22: 0.2189344373189497}
> PL/Python function "madlib_keras_fit_multiple_model
>  
>  



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to