-
NDArray does not support
# A: NDArray for a in A:
only support
# A: NDArray for i in range(len(A))
-
When creating new NDArray, the shape must be a tuple, could not be an int, even if creating a 1D vector, e.g.
extended_symbols = array_api.full((2 * len(target) + 1, ), self.blank, device=target.device) # accepable, (2 * len(target) + 1) false, must end with a ',', (2 * len(target) + 1, ) true
-
NDArray getitem could only create an NDArray with the same shape as before, e.g. A shape (3, 4), A[1, :3] is with shape (1, 3) rather than (3, )
-
If want to reshape a getitem sliced NDArray, must first compact() than reshape target_trunc = target[batch_itr, :int(target_lengths[batch_itr].numpy())].compact().reshape(-1)
-
Only support compact with positive offset
-
a: NDArray([1.0]) != 1.0, cannot directly compare NDArray with number, must use float(a.numpy()) != 1.0
-
Notifications
You must be signed in to change notification settings - Fork 1
Kelton8Z/needleASR
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
About
No description, website, or topics provided.
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published