pytorch/xla

Pull Requests Commits

Merge branch 'master' into 9082-educate-users-on-mat-mul-precision

yaoshiang committed 166 days ago

6538839e

Subset of existing, approved PR (#9204)

yaoshiang committed 166 days ago

Verified c7ef4395

Fixed reloading of sharded master weights (#9224)

avizon-aws committed 166 days ago

Verified c0f1e620

Update the OpenXLA pin (#9216)

haifeng-jin committed 166 days ago

Verified ff57e53c

[ragged-paged-attn] Apply kv mask to filter out NaNs (#9219)

bythew3i committed 166 days ago

Verified 82e7ee80

Document test/neuron/run_tests.sh. (#9205)

zhanyong-wan committed 167 days ago

Verified e28174c9

Improve CI build speed by letting bazel decide how many worker threads to have (#9209)

zhanyong-wan committed 167 days ago

Verified eedc8d23

Fix the examples in API_GUIDE (#9213)

zhanyong-wan committed 167 days ago

Verified 7ce8493d

Reflow ci.md to 80 chars (#9151)

tengyifei committed 167 days ago

Verified 334b4d3f

Update ci.md and fix typo (#9150)

tengyifei committed 167 days ago

Verified 0b6349b0

Clean up env var usages in run_tests.sh scripts. (#9207)

zhanyong-wan committed 167 days ago

Verified 6e3f4076

Split expensive TPU tests to run in parallel (#9198)

ghpvnist committed 167 days ago

Verified b5937198

Make 2 more run_tests.sh support test selection (#9192)

zhanyong-wan committed 168 days ago

Verified 01b5408d

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

zhanyong-wan committed 170 days ago

Verified edc1a882

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

yaochengji committed 171 days ago

Verified f39434ab

get master ip address for neuron device (#9120)

aws-zhenguo committed 171 days ago

Verified fd10b960

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

aws-yyjau committed 171 days ago

Verified 3781920c

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

ghpvnist committed 171 days ago

Verified 8c71b049

Add assume_pure_torch implementation for forward pass only (#9135)

bhavya01 committed 171 days ago

Verified e724225f

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

avizon-aws committed 172 days ago

Verified b8c42523

minor typo, plus adjusting build script.

yaoshiang committed 172 days ago

7bf49b1b

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

ghpvnist committed 172 days ago

Verified bbafe357

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

zhanyong-wan committed 172 days ago

Verified a9ba21cf

Fix a clang compilation error. (#9156)

zhanyong-wan committed 172 days ago

Verified ffd4488c

add aarch64 platform build support (#8663)

snadampal committed 173 days ago

Verified a061a704

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

ghpvnist committed 173 days ago

Verified 05ac08f3

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

zhanyong-wan committed 173 days ago

Verified 1048005b

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

zhanyong-wan committed 173 days ago

Verified 80f64710

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

zhanyong-wan committed 173 days ago

Verified 40810f01

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

zhanyong-wan committed 173 days ago

Verified d9d8f38c

Older