pytorch/xla

Pull Requests Commits

Fix duplicate @unittest.skip test_core_aten_ops.py (#5976)

wonjoo-wj committed 1 year ago

Verified 990c673c

Add script for updating core aten opset issue (#5821)

wonjoo-wj committed 1 year ago

Verified 74fd6e5f

Optimize autocast tests. (#5970)

vanbasten23 committed 1 year ago

Verified 67cb9ca6

Allow benchmark runner config: `openxla` for inference runs. (#5939)

ysiraichi committed 1 year ago

Verified dcdd66e3

Add all-gather coalescing for FSDP/ZeRO1 (#5950)

jeffhataws committed 1 year ago

Verified 12719646

Increase tolerance for tan (#5915)

qihqi committed 1 year ago

Verified 2c4983dd

Enable TF32 multiplication by default. (#5965)

golechwierowicz committed 1 year ago

Verified e98cb665

Vectorize local shard retrieval (#5826)

jonb377 committed 1 year ago

Verified c919973d

use a common random seed for test core aten ops (#5949)

qihqi committed 1 year ago

Verified f6a775c4

Distribute Literal->Tensor copies across thread pool (#5825)

jonb377 committed 1 year ago

Verified ec54fd44

Add stack frame id metadata and correct filename and line number for custom op_name (#5838)

mrnikwaws committed 1 year ago

Verified 5fa89223

Update troubleshooting doc with few common env var combination (#5929)

JackCaoG committed 1 year ago

Verified 6c5b7b8e

Add graph hash and num input/output to PT_XLA_DEBUG (#5947)

JackCaoG committed 1 year ago

Verified f3b75ba1

Refactor ExecuteReplicated to operate on sharded data directly (#5737)

will-cromar committed 1 year ago

Verified 7b92a949

Error when changing `PJRT_DEVICE` after runtime initialized (#5948)

will-cromar committed 1 year ago

Verified 346ef7c1

Add benchmark noise reducing info. (#5924)

golechwierowicz committed 1 year ago

Verified c8ce0620

Add link to work log issue in FIX_LOWERING_FOR_CORE_ATEN_OPS.md (#5944)

wonjoo-wj committed 1 year ago

Verified 09f2a14c

Parallelize d2h transfers (#5824)

jonb377 committed 1 year ago

Verified b9475d9e

Update pjrt.md (#5941)

Liyang90 committed 1 year ago

Verified eb728a89

Truncate python stack when outputting frame that cause the graph executation (#5933)

JackCaoG committed 1 year ago

Verified a5e07389

Don't fallback for pow (#5919)

qihqi committed 1 year ago

Verified 402166be

Lower quant/dequant torch op to StableHLO (#5763)

lsy323 committed 1 year ago

Verified a3b0c6ec

Fix `as_strided` for inputs smaller than the arguments specification. (#5914)

ysiraichi committed 1 year ago

Verified 4ac255ae

Collect CUDA/CPU profiling info into result sheets. (#5921)

golechwierowicz committed 1 year ago

Verified c03afb14

Fix get_gpu_name returning empty name. (#5925)

golechwierowicz committed 1 year ago

Verified 16a40cc6

Revert "Configure PATH and LD_LIBRARY_PATH to include cuda (#5761)" (#5928)

vanbasten23 committed 1 year ago

Verified cfb36175

Copy input tensors before async transfer (#5830)

will-cromar committed 1 year ago

Verified b69fb93d

Refactor Dynamo (custom op) integration code (#5805)

yeounoh committed 1 year ago

Verified 3385bd64

update result analyzer script (#5920)

zpcore committed 1 year ago

Verified 037bbd66

Fix tier import failure for relative import (#5918)

frgossen committed 1 year ago

Verified fe7b0075

Newer Older