Fixed: #12233 #12249
Fixed: #12233 #12249
Conversation
|
There was a problem hiding this comment.
Please see my comment in the conversation
Note that y_true is an array so comparing it to 0 might not be a good idea here, and there are nonzero entries that should be calculated instead of just returning 0.
This causes
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
|
@anandfresh I do not think the PR should be approved as it is not correct and do not resolve what the issue intends. |
@kevin1kevin1k - Yes, my bad ! I didn't see it is an array variable . Ideally he has to check if the array length is greater than 0 |
| @@ -659,7 +659,10 @@ def kullback_leibler_divergence(y_true: np.ndarray, y_pred: np.ndarray) -> float | |||
| if len(y_true) != len(y_pred): | |||
| raise ValueError("Input arrays must have the same length.") | |||
|
|
|||
| kl_loss = y_true * np.log(y_true / y_pred) | |||
| kl_loss = 0 | |||
| if y_true != 0: | |||
There was a problem hiding this comment.
Shouldn't you check the length of the y_true array greater than 0?
|
Closing tests_are_failing PRs to prepare for Hacktoberfest |
I have added a logic that restricts the evaluation of log function if y_true is 0.
Checklist: