pytorch/xla

Pull Requests Commits

move more tests to v4 only

JackCaoG committed 2 years ago

60224207

Move fsdp autowrap test to v4 TPUCI only

JackCaoG committed 2 years ago

65d98a03

[torch_xla2] Implement `bessel_*`, `chebyshev_*`, `hermite_*`, `erfcx`, `i0e` (#7723)

will-cromar committed 2 years ago

Verified 07a83636

add test for dynamo + traceable collectives (#7745)

JackCaoG committed 2 years ago

Verified b9c19b45

[benchmarks] Turn CUDAGraphs on for inductor. (#7749)

ysiraichi committed 2 years ago

Verified 0b990857

[benchmarks] Set matrix multiplication precision. (#7748)

ysiraichi committed 2 years ago

Verified 2870e93b

[benchmarks] Fix batch size logic. (#7747)

ysiraichi committed 2 years ago

Verified 62807012

Support `@xla.step` decorator without parentheses (#7746)

will-cromar committed 2 years ago

Verified 2e1ebfc0

fix TPU test regarding word_size (#7743)

zpcore committed 2 years ago

Verified d77bf6fe

Fix the crash when symint is part of the output with dynamic torch.compile (#7735)

JackCaoG committed 2 years ago

Verified df59b1c3

Use torch_xla.experimental.compile for all examples (#7642)

JackCaoG committed 2 years ago

Verified 735dd435

Fix the SPMD doc strng (#7718)

JackCaoG committed 2 years ago

Verified b4d82588

Update readme with 2.4 release version (#7742)

bhavya01 committed 2 years ago

Verified 2c770683

Rename `aten_cpu_fallback` into `aten_fallback`. (#7646)

ysiraichi committed 2 years ago

Verified 9d113adf

deprecate xrt_world_size (#7679)

zpcore committed 2 years ago

Verified 7f8ef791

Extend timeout for `TpuInfoCliTest.test_single_process_e2e_one_chip` (#7733)

will-cromar committed 2 years ago

Verified fb9f54ed

Optimize dynamo dynamic shape caching (#7726)

JackCaoG committed 2 years ago

Verified 33f69ad8

Do not try to trim the graph (#7725)

JackCaoG committed 2 years ago

Verified a34d28b4

Make CUDA OpenXLA fallback the default. (#7630)

ysiraichi committed 2 years ago

Verified 806de832

Re-implement and refactor the verifier. (#7724)

ysiraichi committed 2 years ago

Verified c69742c0

[torch_xla2] Add decorator to promote int inputs (#7729)

will-cromar committed 2 years ago

Verified c46cdf23

Add final 2.4 trigger (#7732)

bhavya01 committed 2 years ago

Verified 807c595d

[torch_xla2] Opinfo tests for `rsqrt`, `scalar_tensor`, `i0`, `slice*`, `scalar.window.*` (#7716)

will-cromar committed 2 years ago

Verified 2106ea0d

Build CUDA plugin before `torch` and `torch_xla` in nightly build (#7721)

will-cromar committed 2 years ago

Verified 42aa7bdc

Add support for dynamic shape in dynamo (#7676)

wonjoo-wj committed 2 years ago

Verified 2b6b4613

Update contributing doc after walkthrough (#7332)

tengyifei committed 2 years ago

Verified 5ea4535f

Unify the logics to check eager mode (#7709)

JackCaoG committed 2 years ago

Verified 4ba63ffd

Update OpenXLA pin as of 07172024 (#7715)

wonjoo-wj committed 2 years ago

Verified ea848830

Update eager.md (#7710)

JackCaoG committed 2 years ago

Verified 0ddc63a9

Add trapz, trapezoid, div (#7706)

ManfeiBai committed 2 years ago

Verified 2f024274

Older