pytorch xla issues - Githubissues

pytorch / xla

Enabling PyTorch on XLA Devices (e.g. Google TPU)

https://pytorch.org/xla

Other

2.49k stars 483 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Unit test pass

#8416 qihqi opened 3 hours ago
0
[LoweringContext] Modularize for enhanced user control and optimization

#8415 rpsilva-aws opened 12 hours ago
0
[LoweringContext] Support explicit device data parameters for scalar inputs

#8414 rpsilva-aws opened 13 hours ago
0
Review documentation in the docs/source/learn directory

#8413 mikegre-google opened 14 hours ago
0
[Documentation] Added a section pointing readers to the AI-Hypercomputer/tpu-recipies repo for reference model implementations.

#8412 mikegre-google opened 14 hours ago
0
Update XLA pin to 11/22

#8411 JackCaoG closed 13 hours ago
1
``torch.split`` followed by ``torch.cat`` fails to restore the tensor (v2.1-2.5)

#8410 jeffhataws opened 3 days ago
1
Switch to libtpu package from the libtpu-wheels registry

#8409 tengyifei opened 3 days ago
2
From passing zero tests to 1,153 out of an unknown number of tests for nn.functional.max_pool2d

#8408 hosseinsarshar closed 3 days ago
2
Dataloading hangs on Trillium when using recent whls

#8407 miladm opened 4 days ago
1
Getting "undefined symbol: _ZN5torch4lazy13MetricFnValueB5cxx11E" with torch-xla nightly wheel for 2.6

#8406 jeffhataws opened 4 days ago
2
Einsum is not added to the supported list for autocast

#8405 avizon-aws opened 4 days ago
2
Force specialize_float = True for torch_xla

#8404 ezyang closed 4 days ago
3
[torch_xla2] Update test_ops.py

#8403 ManfeiBai opened 5 days ago
0
Kaggle Notebook: model return loss None on TPU

#8402 manh3152924 opened 5 days ago
1
Disable the paged_attention test to unblock head.

#8401 vanbasten23 closed 5 days ago
0
Add an example that train the torchtitan version of llama.

#8400 qihqi closed 3 days ago
0
Add multiple TPU generations to the TPU CI tests

#8399 miladm opened 6 days ago
0
Migrate from capture_pre_autograd_graph to export_for_training

#8398 yushangdi opened 6 days ago
1
Update ARC runner version to 2.321.0

#8397 mbzomowski closed 6 days ago
0
DDP doesn't synchronize weights

#8396 darudimimi opened 6 days ago
0
torch.layer_norm causes RESOURCE_EXHAUSTED

#8395 radna0 opened 6 days ago
0
[torch_xla2] Create jtorchvision_roi_align.py

#8394 ManfeiBai opened 6 days ago
0
Only write to HBM at the last iteration.

#8393 vanbasten23 closed 5 days ago
5
Update XLA pin to 11/17

#8392 JackCaoG closed 1 week ago
0
Unify how cpp test is being run

#8391 JackCaoG closed 6 days ago
0
[TPU][torch.compile] How to introduce in-place custome Ops through Pallas ?

#8390 xinli-centml closed 1 week ago
1
Prepare a subsection to educate users on the PyTorch workloads on AI-Hypercomputer

#8389 miladm opened 1 week ago
0
Need help validating TPU/XLA devices support for ComfyUI.

#8388 radna0 opened 1 week ago
2
Can Triton be used with XLA/TPU devices?

#8387 radna0 opened 1 week ago
1
Fix the fallback op in SPMD

#8386 JackCaoG closed 1 week ago
0
How to write in-place custom ops compatible with torch.compile using pallas

#8385 soodoshll opened 1 week ago
6
norm and pca_lowrank op info test from issue #7528

#8384 vyom1611 opened 1 week ago
0
update the path to tsl/protobuf/error_codes.pb.h

#8383 JackCaoG closed 1 week ago
1
Fix incorrect xla type for view_as_complex

#8382 JackCaoG closed 1 week ago
1
Support splitting physical axis in HybridMesh

#8381 tengyifei opened 1 week ago
0
How are PJRT asynchronous executions throttled by torch_xla?

#8380 mcuiaws opened 1 week ago
7
Confusing text in bazel.md

#8379 mikegre-google opened 1 week ago
2
[torch_xla2] add `nn.functional.multilabel_margin_loss` ... `randint`

#8378 ManfeiBai closed 1 week ago
0
Missing python 3.9 and 3.11 wheels for torch-xla 2.5.1 on pypi

#8377 jeffhataws closed 1 week ago
3
Fix op info test for svd_lowrank

#8376 dvhg closed 1 week ago
0
Revert S64 and U64 type change for Neuron device due to hardcoded dtype checks in torch-xla

#8375 aws-nm9 closed 1 week ago
1
Remove unnecessary prints in PagedAttention

#8374 WoosukKwon closed 1 week ago
1
int8 StableHLO export

#8373 Wheest opened 1 week ago
3
The precision error when xla parses irrational numbers leads to inconsistent calculation results with torch cuda

#8372 mars1248 opened 1 week ago
1
TPU Trillium Base Docker Image cannot initialize

#8371 hsebik opened 1 week ago
7
Fix XLA type of the view_as_real

#8370 JackCaoG closed 1 week ago
0
Improve some typing annotations

#8369 tengyifei closed 2 weeks ago
0
[torch_xla2] Enable nn.functional.cosine_embedding_loss

#8368 ManfeiBai closed 1 week ago
0
xla bf16 compute accuracy issues

#8367 mars1248 opened 2 weeks ago
0