You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Allow different input types for DataStats transform by default #6844
Is your feature request related to a problem? Please describe.
Currently DataStats supports Ndarray or Tensor and breaks by default for any other data type if the shape attribute is not available.
There are four items for DataStats and in the default behavior,
data_type: Works irrespective of the provided type.
data_shape: Breaks when shape attribute is missing
value_range: Works irrespective of provided type as there is a conditional check for type
data_value: Off by default
Describe the solution you'd like
Could we streamline 2 and 3 so that there is a conditional check for the data_shape similar to type checks for value_range?
Activity
add shape check for arbitrary types for DataStats
add shape check for arbitrary types for DataStats (#7082)