pytorch/xla

Pull Requests Commits

Merge benchmark scripts into the main branch (#5913)

frgossen committed 1 year ago

Verified 94ca4597

Create FIX_LOWERING_FOR_CORE_ATEN_OPS.md (#5909)

wonjoo-wj committed 1 year ago

Verified 3fa035e3

Re-implement `ComputationClient::WaitDeviceOps` (#5811)

will-cromar committed 1 year ago

Verified 175e86b8

Add python 3.9 and 3.11 build for nightly (#5840)

JackCaoG committed 1 year ago

Verified 57e60352

Add a unittest testing each core ATen ops. (#5796)

qihqi committed 1 year ago

Verified 2b04794a

Sync `xla_args` before computation. (#5823)

ysiraichi committed 1 year ago

Verified f80e3ba9

Use cuda 12.1 and build pytorch with cuda enabled. (#5831)

vanbasten23 committed 1 year ago

Verified 8c03ac96

Configure PATH and LD_LIBRARY_PATH to include cuda (#5761)

vanbasten23 committed 1 year ago

Verified 4fb54281

Extend HLO metadata to include class hierarchy information (#5715)

mrnikwaws committed 1 year ago

Verified 5e503794

[SPMD] fix bug with XLAShardedTensor.__repr__ (#5807)

yeounoh committed 1 year ago

Verified 6addbde3

Use upstream XLA concurrency utilities (#5799)

will-cromar committed 1 year ago

Verified 6e661304

Add python 3.11 trigger for 2.1 release build (#5810)

JackCaoG committed 1 year ago

Verified a3bbbb1e

Enable passing down dynamic dimensions from torch to XLA (#5790)

lsy323 committed 1 year ago

Verified a0a048e1

Don't set $TPU_LIBRARY_PATH during import

will-cromar committed 1 year ago

Verified d8c6a83f

Record the lazy tracing time(C++) in metrics (#5757)

JackCaoG committed 1 year ago

Verified 58ce7e17

Use TPU profiler plugin (#5793)

will-cromar committed 1 year ago

Verified 3b34cb27

Add GKE support and various usability improvements in CheckpointManager (#5770)

jonb377 committed 1 year ago

Verified 5710a83c

Lower AtenFull op (#5781)

danielvegamyhre committed 1 year ago

Verified b3ed82c9

Transfer data directly to the device (#5772)

will-cromar committed 1 year ago

Verified 05a3cdda

[SPMD] suppor DTensor API integration (#5776)

yeounoh committed 1 year ago

Verified e60428d1

disable test_set temporarily (#5792)

yeounoh committed 1 year ago

Verified 6f4183c8

Make the pjrt gpu allocator configurable (#5759)

anw90 committed 1 year ago

Verified 56733fbc

Remove some unused code from `csrc/runtime` (#5785)

will-cromar committed 1 year ago

Verified 79557ccd

Implement mark_sharding as a custom op to support dynamo spmd activation sharding (#5712)

wonjoo-wj committed 1 year ago

Verified 367f47f5

Use default method in stablehlo txt/bytecode getter of StableHLOGraphModule (#5745)

lsy323 committed 1 year ago

Verified d2f389e7

Dsiable C++ test correctly (#5783)

JackCaoG committed 1 year ago

Verified d5d02306

Refactor type conversion functions out of `//torch_xla/csrc:tensor` (#5777)

will-cromar committed 1 year ago

Verified ee210bb7

Disable flaky cpp test (#5779)

JackCaoG committed 1 year ago

Verified 963204a3

fixing neuron import (#5775)

aws-kingrj committed 1 year ago

Verified bd29e796

Support autocheckpointing (#5753)

jonb377 committed 1 year ago

Verified 46643801

Newer Older