pytorch/xla

Pull Requests Commits

Asymmetric quantized matmul support (#7626)

lsy323 committed 1 year ago

Verified 289471cb

Easier way to make img show up in our doc page (#7644)

JackCaoG committed 1 year ago

Verified 72069750

Move tensor constructors to `jaten` (#7615)

will-cromar committed 1 year ago

Verified 5e9da938

[torch_xla2] Add broadcast_to, isin, item, lcm, lerp (#7638)

ManfeiBai committed 1 year ago

Verified ccd3da2c

[torch_xla2] Add fmax, fmin, frexp (#7603)

ManfeiBai committed 1 year ago

Verified c62b1d46

Add blockwise quantized dot support (#7605)

lsy323 committed 1 year ago

Verified 88bcb45f

Update `apt-get update` in upstream image build (#7640)

will-cromar committed 1 year ago

Verified edd24532

separate PyTorch/XLA docs (#7593)

JackCaoG committed 1 year ago

Verified 275bf919

Update test_pallas_spmd.py for CI (#7637)

bhavya01 committed 1 year ago

Verified c84c8932

Fallback `_embedding_bag_backward` and force `sparse=false`. (#7584)

ysiraichi committed 1 year ago

Verified 3bcb1fb1

fix eager mode spmd module loading with fsdpv2 (#7631)

JackCaoG committed 1 year ago

Verified 10a51300

Update test_pallas.py for CI (#7629)

ManfeiBai committed 1 year ago

Verified 05a6487f

Introduce CUDA OpenXLA fallback. (#7318)

ysiraichi committed 1 year ago

Verified c782e0de

Early exit for clip_grad_norm_ (#7621)

alanwaketan committed 1 year ago

Verified 172097b6

[Pallas] Improve the user guide (#7625)

alanwaketan committed 1 year ago

Verified 661f1914

SPMD parameter wrapping fix review comment (#7619)

JackCaoG committed 1 year ago

Verified 3ce6ffb2

add tpu example for torchrun on GKE (#7620)

chengcongdu committed 1 year ago

Verified 51f89c85

Update nightly package version to 2.5 (#7623)

bhavya01 committed 1 year ago

Verified bbf0db79

added how to tell when ops need to be lowered (#7624)

nhira committed 1 year ago

Verified 2c011379

Update ansible config to separately update submodules when torch_pin is specified (#7617)

bhavya01 committed 1 year ago

Verified 59c93619

Handle the parameter wrapping for SPMD (#7604)

JackCaoG committed 1 year ago

Verified 4b7e518d

Lower empty_like, empty_permute, empty_strided (#7610)

lsy323 committed 1 year ago

Verified 1ec9f2bb

Fix GPU yaml in kubernetes.md (#7612)

will-cromar committed 1 year ago

Verified 7ec57784

[torch_xla2] Add lowerings for `rand_like` and `randn_like`, and enable some opinfo tests (#7170)

will-cromar committed 1 year ago

Verified 8f5de773

Run paged_attention test with megacore mode only on TPU v4 (#7609)

bhavya01 committed 1 year ago

Verified e8cd9855

Reland: `mul`: convert inputs to result type. (#7602)

ysiraichi committed 1 year ago

Verified 77f66b51

[benchmarks] Fix `moco` initialization device. (#7598)

ysiraichi committed 1 year ago

Verified 86169d27

Add documentation for Triton kernels in PTXLA (#7595)

bhavya01 committed 1 year ago

Verified e7a8700e

Update JAX version to fix TPU CI (#7601)

will-cromar committed 1 year ago

Verified 593dce8b

add spmd gpu doc back (#7592)

JackCaoG committed 1 year ago

Verified 755ae846

Newer Older