pytorch/xla

Pull Requests Commits

add several examples testing the failure modes of tensor parallel

bfolie committed 1 year ago

74922eca

add scripts investigating send/recv and XLA ops on sharded tensors

bfolie committed 1 year ago

a4203d56

Merge branch 'master' of https://github.com/pytorch/xla into bfolie/distributed-trials

bfolie committed 1 year ago

b9c5ff7a

Allow run_tests to run a subset of the tests. (#9190)

zhanyong-wan committed 1 year ago

Verified edc1a882

[Kernel] Use call_jax to simplify the gmm pallas kernel wrapper (#9180)

yaochengji committed 1 year ago

Verified f39434ab

get master ip address for neuron device (#9120)

aws-zhenguo committed 1 year ago

Verified fd10b960

test: add test_xla_graph_execution to test flags (_set_allow_execution with PT_XLA_DEBUG_LEVEL) (#9171)

aws-yyjau committed 1 year ago

Verified 3781920c

Replace `xm.mark_step` with `torch_xla.sync()` wherever possible (#9070)

ghpvnist committed 1 year ago

Verified 8c71b049

Merge branch 'master' of https://github.com/pytorch/xla into bfolie/distributed-trials

bfolie committed 1 year ago

0d844c69

make new directory for test scripts, add script to run all_gather on smaller process group

bfolie committed 1 year ago

a6235f61

Add assume_pure_torch implementation for forward pass only (#9135)

bhavya01 committed 1 year ago

Verified e724225f

fix: reload the param groups in optimizer (#9164)

avizon-aws committed 1 year ago

Verified b8c42523

Replace implementation of `xla_torch.sync()` to `xm.mark_step()` (#9086)

ghpvnist committed 1 year ago

Verified bbafe357

Switch bazel default spawn strategy to `sandboxed` (#9168)

zhanyong-wan committed 1 year ago

Verified a9ba21cf

Fix a clang compilation error. (#9156)

zhanyong-wan committed 1 year ago

Verified ffd4488c

add aarch64 platform build support (#8663)

snadampal committed 1 year ago

Verified a061a704

Don't install wheels when building docs (#9153)

ghpvnist committed 1 year ago

Verified 05ac08f3

Update XLA to a 2025/5/13 revision. (#9155)

zhanyong-wan committed 1 year ago

Verified 1048005b

Add instructions on how to rebase local commits (#9162)

zhanyong-wan committed 1 year ago

Verified 80f64710

Add a linter check to prevent *.cc file names. (#9147)

zhanyong-wan committed 1 year ago

Verified 40810f01

Rename *.cc files to *.cpp for consistency. (#9149)

zhanyong-wan committed 1 year ago

Verified d9d8f38c

write script to lower a collective op and print output

bfolie committed 1 year ago

175a9970

add traceback to verbose output

bfolie committed 1 year ago

79273f3a

Update instructions for updating forked pytorch (#9154)

zhanyong-wan committed 1 year ago

Verified 1f9dd8f3

Re-land: Add `clang` link to PATH. (#9144)

ysiraichi committed 1 year ago

Verified dfac6042

Ensure that all of PyTorch/XLA C++ is compiled with exceptions enabled. (#9146)

zhanyong-wan committed 1 year ago

Verified 8450607a

Test torchax on Python 3.10 - 3.12 (#9139)

tengyifei committed 1 year ago

Verified 259b438e

Silence distributed warning (#9140)

tengyifei committed 1 year ago

Verified 99e2d3c9

[benchmarks] Fix run single config command error (#9115)

haifeng-jin committed 1 year ago

Verified d2ab82d5

Refine the gradient accumulation API (#9078)

rpsilva-aws committed 1 year ago

Verified 7d3d6e95

Older