Skip to content

Add warning about inactive NJT development to related docs #3726

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
jbschlosser wants to merge 5 commits into
base: main
Choose a base branch
from
+6 −2
Open

Add warning about inactive NJT development to related docs #3726

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
370b06c
Add warning about inactive NJT development to related docs
jbschlosser Jan 16, 2026
396a875
Use RST syntax
jbschlosser Jan 16, 2026
eec5911
Update unstable_source/nestedtensor.py
svekars Jan 16, 2026
9397cc9
Update nestedtensor.py
svekars Jan 16, 2026
a7a73e1
Merge branch 'main' into njt_warning
svekars Jan 16, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions intermediate_source/transformer_building_blocks.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -79,6 +79,9 @@
# sequence lengths. They eliminate the need for the bug-prone practices of explicit
# padding and masking (think ``key_padding_mask`` in ``nn.MultiHeadAttention``).
#
# .. warning::
# Nested tensors are not currently under active development. Use at your own risk.
#
# * `scaled_dot_product_attention <https://pytorch.org/tutorials/intermediate/scaled_dot_product_attention_tutorial.html>`_
#
# ``scaled_dot_product_attention`` is a primitive for
Expand Down
5 changes: 3 additions & 2 deletions unstable_source/nestedtensor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,9 @@
Getting Started with Nested Tensors
===============================================================
.. warning::
Nested tensors are not currently under active development. Use at your own risk.
Nested tensors generalize the shape of regular dense tensors, allowing for representation
of ragged-sized data.
Expand All @@ -21,8 +24,6 @@
they are invaluable for building transformers that can efficiently operate on ragged sequential
inputs. Below, we present an implementation of multi-head attention using nested tensors that,
combined usage of ``torch.compile``, out-performs operating naively on tensors with padding.
Nested tensors are currently a prototype feature and are subject to change.
"""

import numpy as np
Expand Down