pytorch/xla

Pull Requests Commits

Add python test for sizeAdd node and implement `int_` for XLASymNodeImpl. (#4218)

vanbasten23 committed 2 years ago

Verified f790bc8a

Introduce virtual device class (#4091)

steventk-g committed 2 years ago

Verified b2bd721c

Replace XLA_COUNTER with TORCH_LAZY_COUNTER (#4208)

alanwaketan committed 2 years ago

Verified 788001c5

Improve XLASynNodeImpl str() format (#4216)

vanbasten23 committed 2 years ago

Verified a46c616e

Add a api to reset and extract RNG Seed (#4196)

JackCaoG committed 2 years ago

Verified 0daf6fe9

Implement `xm.rendezvous` with XLA collective communication for PJRT (#4181)

will-cromar committed 2 years ago

Verified a9eb3459

[SPMD] fix sharding util bugs & implement unit test cases for 5D tile sharding (#4192)

yeounoh committed 2 years ago

Verified 95fcf680

Run cached graph api should grab device lock first (#4195)

JackCaoG committed 2 years ago

Verified 08a9c5dd

Add the dynamic shape unit test the CI. (#4180)

vanbasten23 committed 2 years ago

Verified a843d8bf

Create new Cloud Build workflow for nightly images/packages (#4188)

will-cromar committed 2 years ago

Verified 3f102965

[LTC] Adopt getBackend()->GetComputationDataFromNode() (#4191)

alanwaketan committed 2 years ago

Verified dd9b67ff

Fix multi-gpu metrics report (#4187)

ymwangg committed 2 years ago

Verified 44362e6d

[LTC] Use getBackend()->ExecuteComputation() (#4186)

alanwaketan committed 2 years ago

Verified ee65ff15

[LTC] Use torch::lazy::Util::ComputePostOrder (#4178)

alanwaketan committed 2 years ago

Verified ae35e1e2

unblock dynamo integration, get autograd ops to work with python tensor subclasses (#4165)

bdhirsh committed 2 years ago

Verified 4a0aaefb

[WIP] Partially Codegen Atan2 (#4184)

ManfeiBai committed 2 years ago

Verified b5496122

Fix check_tensor_need_materialization (#4185)

JackCaoG committed 2 years ago

Verified b1f167e6

[LTC] Make some torch::lazy::Node* to const T (#4182)

alanwaketan committed 2 years ago

Verified 0115cc00

Additional dynamo test cases (#4175)

JackCaoG committed 2 years ago

Verified bc2694ff

Make ExecuteComputationOptions as default (#4177)

alanwaketan committed 2 years ago

Verified a01b9999

[LTC] Employ more Computation (#4176)

alanwaketan committed 2 years ago

Verified 834e3efd

Remove upsample_*_backward.vec ops (#4087)

wonjoo-wj committed 2 years ago

Verified f4dd9906

[LTC] Use upstream DeviceDataInfo (#4172)

alanwaketan committed 2 years ago

Verified 54f24e25

[SPMD] Mesh to support custom device order. (#4162)

yeounoh committed 2 years ago

Verified b096c5cf

[LTC] Remove XLAIrBuilder::MakeView (#4154)

alanwaketan committed 2 years ago

Verified f7c0775b

Companion PR for 'SymIntify _copy functionalization kernels (and _copy_out too)' (#4164)

ezyang committed 2 years ago

Verified 79cbdb60

Add some basic Dynamo test (#4159)

JackCaoG committed 2 years ago

Verified 515d53e9

Enable `padded` option for even tensor sharding (#4156)

yeounoh committed 2 years ago

Verified 5ff192ea

Update codegen to use xla::Shape (#4111)

wonjoo-wj committed 2 years ago

Verified 587a6b6d

Refactor argparser to be consistent with torch.distributed (#4149)

muellerzr committed 3 years ago

Verified bde1bc6a

Newer Older