| Column | Data Type | Description |
|---|---|---|
| accumulate_column | Same as in InputTable | One column each for the column names specified in accumulate_column. Use an asterisk to specify all columns. |
| output_column | Coincides with matching OutputFormat | If a simple model output tensor name is used in the map and the matching OutputFormat is not an array type, a single column is added to the output |
| prefixed_output_column | Coincides with matching OutputFormat | If a simple model output tensor name is used in the map and the matching OutputFormat is an an array type, multiple columns are added to the output prefixed with the column name and added '_0', '_1','_2' values to the new columns. |
| assignment_output_column(simple) | Coincides with matching OutputFormat | If a model output tensor name is used in the map followed by an equal sign and a single column name and the matching OutputFormat is not an array type, a single column is added to the output. |
| assignment_output_columns(prefixed) | Coincides with matching OutputFormat | If a model output tensor name is used in the map followed by an equal sign and a single column name and the matching OutputFormat is an an array type, multiple columns are added to the output prefixed with the column name and added '_0', '_1','_2' values to the new columns. |
| assignment_output_columns(custom named) | Coincides with matching OutputFormat | If a model output tensor name is used in the map followed by an equal sign and a list of column names and the matching OutputFormat is an an array type, multiple columns are added to the output with the column names specified in the list. |
| Softmax_output(simple) | Coincides with matching OutputFormat | If a model output tensor name is enclosed in a softmax function signature and is followed by an equal sign and a single column name and the matching OutputFormat is not an array type, a single softmax computed output column is added. |
| Softmax_output(prefixed) | Coincides with matching OutputFormat | If a model output tensor name is enclosed in a softmax function signature and is followed by an equal sign and a single column name and the matching OutputFormat is an array type, multiple prefixed columns are produced with softmax computed. |
| Softmax_output(custom named) | Coincides with matching OutputFormat | If a model output tensor name is enclosed in a softmax function signature and is followed by an equal sign and a list of column names and the matching OutputFormat is an array type, multiple columns are produced with softmax computed. |
| Argmax_output(prefixed) | Coincides with matching OutputFormat | If a model output tensor name is enclosed in an argmax function signature and is followed by an equal sign and a single column name, two columns are added. One column for the index and one column for the value, both columns prefixed by the single column name. |
| Argmax_output(named) | Coincides with matching OutputFormat | If a model output tensor name is enclosed in an argmax function signature and is followed by an equal sign and a list of two column names (one for the index and one for the value), two columns are added with their respective names. The first column for the index value and the second column for the value at that index. |
| Argmax_Softmax(combo output) | Coincides with matching OutputFormat | If a model output tensor name is enclosed in a softmax function signature which is itself enclosed in an argmax function signature, it behaves exactly as the two Argmax examples above but the output is based on computed softmax values. |
| ModelProperties | String | Input model properties. |