Github user edespino commented on the issue:
https://github.com/apache/incubator-madlib/pull/149
Not sure what this is. While playing around with the module, I accidentally
hit a `nan` condition. Here is the output of the session:
```
madlibtest=# CREATE TABLE iris_data(
madlibtest(# id integer,
madlibtest(# attributes numeric[],
madlibtest(# class_text varchar,
madlibtest(# class integer
madlibtest(# );
CREATE TABLE
madlibtest=# INSERT INTO iris_data VALUES
madlibtest-# (1,ARRAY[5.1,3.5,1.4,0.2],'Iris-setosa',1);
INSERT 0 1
madlibtest=# SELECT madlib.mlp_classification(
madlibtest(# 'iris_data', -- Source table
madlibtest(# 'mlp_model', -- Destination table
madlibtest(# 'attributes', -- Input features
madlibtest(# 'class_text', -- Label
madlibtest(# ARRAY[5], -- Number of units per layer
madlibtest(# 'step_size=0.003,
madlibtest'# n_iterations=5000,
madlibtest'# tolerance=0', -- Optimizer params
madlibtest(# 'tanh'); -- Activation function
WARNING: loss: nan
ERROR: spiexceptions.UndefinedColumn: column "nan" does not exist
LINE 4: ...7731e-05,-0.000170379656714,0.000110138191401,1.0,nan,1.0,1....
^
QUERY:
SELECT madlib.internal_mlp_igd_distance(
(NULL)::DOUBLE PRECISION[],
ARRAY[
2.0,4.0,5.0,1.0,0.003,1.0,2.0,1.0,1.0,1.0,1.0,1.0,-0.000161677781273,2.34991541167e-05,-5.27816928274e-05,1.6207561377e-06,9.30167921009e-05,-2.2322484638e-06,-8.8803764219e-05,6.92299455941e-05,0.000174430298478,1.36067005437e-07,-0.000126402572755,0.00010846859258,-0.000131999662994,-0.000111939884066,0.000154233933791,9.62727960033e-05,7.92279598497e-05,3.55520197415e-05,4.28280053118e-05,-0.000136356574146,-3.07769363501e-05,-8.97131375537e-05,-0.000117176292437,-9.89281636713e-05,0.000234615355418,1.0,1.0,1.0,1.0,1.0,1.0,2.00514836476e-05,-9.37533114569e-05,7.58698692162e-05,-2.64470277731e-05,-0.000170379656714,0.000110138191401,1.0,nan,1.0,1.0,1.0,1.0,1.0,-0.000161677781273,2.34991541167e-05,-5.27816928274e-05,1.6207561377e-06,9.30167921009e-05,-2.2322484638e-06,-8.8803764219e-05,6.92299455941e-05,0.000174430298478,1.36067005437e-07,-0.000126402572755,0.00010846859258,-0.000131999662994,-0.000111939884066,0.000154233933791,9.62727960033e-05,7
.92279598497e-05,3.55520197415e-05,4.28280053118e-05,-0.000136356574146,-3.07769363501e-05,-8.97131375537e-05,-0.000117176292437,-9.89281636713e-05,0.000234615355418,1.0,1.0,1.0,1.0,1.0,1.0,2.00514836476e-05,-9.37533114569e-05,7.58698692162e-05,-2.64470277731e-05,-0.000170379656714,0.000110138191401,0.0,0.0
]::double precision[]) as state_dist
CONTEXT: Traceback (most recent call last):
PL/Python function "mlp_classification", line 32, in <module>
True
PL/Python function "mlp_classification", line 140, in mlp
PL/Python function "mlp_classification"
madlibtest=#
```
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
