Skip to content

Issues: pytorch/xla

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

657 Open 1,850 Closed
657 Open 1,850 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

torch_xla.experimental.custom_kernel.flash_attention output does not match F.scaled_dot_product_attention on TPU
#8869 opened Mar 21, 2025 by NickLucche
Nightly CUDA builds are failing CI CI related change xla:gpu
#8868 opened Mar 20, 2025 by tengyifei
@ysiraichi
1
Replace xm.mark_step with torch_xla.sync() in examples and tests documentation enhancement New feature or request usability Bugs/features related to improving the usability of PyTorch/XLA
#8862 opened Mar 19, 2025 by tengyifei
Document the difference between device= vs .to(device) documentation enhancement New feature or request
#8861 opened Mar 19, 2025 by tengyifei
Replace upstream PyTorch GRU module enhancement New feature or request lowering ATen Operation lowering tracing Lazy Tensor tracing
#8860 opened Mar 19, 2025 by tengyifei
Improve torch_xla.compile documentation documentation enhancement New feature or request
#8859 opened Mar 19, 2025 by tengyifei
Document the difference between tracing time and execution time documentation enhancement New feature or request
#8858 opened Mar 19, 2025 by tengyifei
torch.distributed.all_reduce not converted to stableHLO bug Something isn't working SPMD / Distributed stablehlo StableHLO related work
#8854 opened Mar 19, 2025 by AleksKnezevic
1
Have documentation to point to all our environment variables and their meaning documentation usability Bugs/features related to improving the usability of PyTorch/XLA
#8853 opened Mar 19, 2025 by miladm
@bhavya01 @mikegre-google
1
Add a 2-slice pallas training test in pre-submit CI testing Testing and coverage related issues. xla:tpu TPU specific issues and PRs
#8850 opened Mar 18, 2025 by tengyifei
How to compile torch-xla form source? build Build process related matters (e.g. build system). question
#8847 opened Mar 18, 2025 by south-ocean
10
Need a documentation page that always hosts the latest stable documentation documentation enhancement New feature or request
#8846 opened Mar 18, 2025 by tengyifei
@tengyifei
3
Prototype SPMD using local process data enhancement New feature or request SPMD / Distributed
#8842 opened Mar 17, 2025 by pgmoka
@pgmoka
test_ddp_correctness_with_gradient_as_bucket_view fails for multi-device CUDA bug Something isn't working xla:gpu
#8841 opened Mar 17, 2025 by ysiraichi
xs.make_mesh enhancement New feature or request
#8838 opened Mar 14, 2025 by tengyifei
@tengyifei
Inconsistent output when mixing eager with torch.compile bug Something isn't working dynamo
#8832 opened Mar 13, 2025 by liangfu
1
Ship torchax inside torch_xla enhancement New feature or request torchxla2
#8831 opened Mar 13, 2025 by tengyifei
@tengyifei
1
2.7 backport PR request list backport_2.7
#8829 opened Mar 13, 2025 by zpcore
@zpcore
1
Remove non-parallel approaches to get a device enhancement New feature or request usability Bugs/features related to improving the usability of PyTorch/XLA
#8827 opened Mar 12, 2025 by yaoshiang
1
Train loop takes exponentially longer as number of layers increase performance
#8824 opened Mar 12, 2025 by rplsbo
5
CollectivePermute support enhancement New feature or request SPMD / Distributed
#8815 opened Mar 11, 2025 by rpsilva-aws
3
xla device precision issue pytorch divergence XLA behavior doesn't match Pytorch eager frontend
#8811 opened Mar 10, 2025 by mars1248
8
Use jax autograd from PyTorch enhancement New feature or request torchxla2
#8805 opened Mar 7, 2025 by tengyifei
@tengyifei
Re-enable CPU test test/test_python_ops.py -k TestPythonOps for uint8 dtype bug Something isn't working libtpu
#8799 opened Mar 5, 2025 by tengyifei
xb.Op.scalar maps unsigned int to torch signed int bug Something isn't working good first issue Good for newcomers needs reproduction
#8798 opened Mar 5, 2025 by tengyifei
5
ProTip! Adding no:label will show everything without a label.