pytorch/xla

Pull Requests Commits

Move where clear pending IR is called to avoid crash (#5552)

JackCaoG committed 2 years ago

ef34a315

Make cpu tensor on XLA dynamo backend a warning instead of error (#5549) (#5576)

JackCaoG committed 2 years ago

Verified c55ef0b6

Handle dynamo function without input (#5565) (#5577)

JackCaoG committed 2 years ago

Verified f0f9126b

Update torch pin (#5575)

wonjoo-wj committed 2 years ago

Verified 02858b90

Cherry pick `pjrt://` init method rename and doc updates (#5562)

will-cromar committed 2 years ago

Verified ee723327

[Pin Update] Version 20230826(#5527) (#5555)

ManfeiBai committed 2 years ago

Verified 2c07df94

Add resnet50-weight-quant colab notebook (#5407) (#5556)

wonjoo-wj committed 2 years ago

Verified 9bcf3f74

Update r2.1 wheel to be compatible with PyPI (#5550)

will-cromar committed 2 years ago

Verified 691838fb

Fix FSDP for Models with Frozen Weights (#5484) (#5539)

wonjoo-wj committed 2 years ago

Verified 89574d2d

Revert 2.1 terraform changes (#5537)

wonjoo-wj committed 2 years ago

Verified a1d3651f

Add option to unbundle libtpu (#5534) (#5536)

will-cromar committed 2 years ago

Verified 7a01093d

Enable megacore_dense by default (#5520) (#5531)

wonjoo-wj committed 2 years ago

Verified bd1958f3

Update pytorch git tag for r2.1 (#5529)

wonjoo-wj committed 2 years ago

Verified 3b34ae15

Disable cxx abi in ansible when building pt/xla for branch r2.0 (#5332)

vanbasten23 committed 2 years ago

3ac818ec

[SPMD] Propagate replicated output (#5508)

alanwaketan committed 2 years ago

Verified 5c4ce2a9

mark_sharding over a replicated tensor is allowed. (#5513)

yeounoh committed 2 years ago

Verified 6e90569d

Get original_traced_args as example_inputs. (#5511)

qihqi committed 2 years ago

Verified 33f1cd29

Update dynamo.md (#5509)

shauheen committed 2 years ago

Verified 808015d8

Update dynamo doc (#5506)

JackCaoG committed 2 years ago

Verified 2f4ef46d

Fix tupled partition spec test on v3 (#5503)

jonb377 committed 2 years ago

Verified 81063fb8

remove torchvision in stablehlo.py (#5501)

lsy323 committed 2 years ago

Verified 1c42ba74

Add explict error message when tensor is on CPU for dynamo backend (#5499)

JackCaoG committed 2 years ago

Verified 9557fb72

Add a API to explictly init runtime (#5500)

JackCaoG committed 2 years ago

Verified e1ee592d

Support tuples in partition spec (#5488)

jonb377 committed 2 years ago

Verified 7a63c276

Refactor stablehlo API and put them in official location. (#5493)

qihqi committed 2 years ago

Verified 17abce76

Skip a couple tests on TPU due to precision issue (#5496)

JackCaoG committed 2 years ago

Verified d3d6ca30

Always use ExecuteReplicated with SPMD (#5494)

jonb377 committed 2 years ago

Verified 03991d44

Support input sharding changed after first dynamo tracing (#5477)

JackCaoG committed 2 years ago

Verified d1212e90

[SPMD] Make IR sharding custom sharding op (#5433)

alanwaketan committed 2 years ago

Verified dea9b296

[original author: mrnikwaws] Neuron operator support (#5471)

aws-kingrj committed 2 years ago

Verified 3e370b50

Older