pytorch/xla

Pull Requests Commits

add test to the script

JackCaoG committed 2 years ago

7c969741

JackCaoG committed 2 years ago

277bef3d

JackCaoG committed 2 years ago

6dd06232

handle dynamo and everything else

JackCaoG committed 2 years ago

659f1a00

add execution cause

JackCaoG committed 2 years ago

e2748dad

Only master process should print the execution frame info

JackCaoG committed 2 years ago

b85dc013

minor format tweak

JackCaoG committed 2 years ago

6fb077c1

Initial commit for debugging tool

JackCaoG committed 2 years ago

fa407d86

Add information about on-going DTensor API in spmd.md (#5735)

yeounoh committed 2 years ago

Verified 63ea76cd

Add doc for multinode GPU training. (#5704)

vanbasten23 committed 2 years ago

Verified b0dca120

Add Python 3.9 build for 2.1 release (#5744)

will-cromar committed 2 years ago

Verified 421ba3e4

Mangle root scope TF variable name during tf.saved_model export (#5738)

lsy323 committed 2 years ago

Verified 33ff3c29

Merge `--pjrt_distributed` flag with `--ddp` flag. (#5732)

will-cromar committed 2 years ago

Verified 8072a446

Set --xla_latency_hiding_scheduler_rerun to 1 (#5736)

alanwaketan committed 2 years ago

Verified 47a33d0b

Revert "Don't set $TPU_LIBRARY_PATH during import (#5698)" (#5731)

alanwaketan committed 2 years ago

Verified 4baef3c6

Promote in convolution (#5727)

qihqi committed 2 years ago

Verified 47bf2eac

Fix the missing parameter error when running mp_imagenet with torchrun (#5729)

vanbasten23 committed 2 years ago

Verified 294610a1

Make the git version appears in pip (#5728)

alanwaketan committed 2 years ago

Verified ecc0e230

Ensure dist runtime is in sync before shutting down. (#5714)

vanbasten23 committed 2 years ago

Verified fa7828f5

Patch RNG for better memory utilization in dropout layers (#5710)

yeounoh committed 2 years ago

Verified 8f121c31

add op lowering for _prelu_kernel_backward (#5724)

zpcore committed 2 years ago

Verified d28b30d2

Make sure printing XLA tensor only execute the HLO once (#5721)

JackCaoG committed 2 years ago

Verified 1ad310b7

Update OpenXLA pin to 20231022 (#5720)

alanwaketan committed 2 years ago

Verified 43cd6b50

When exporting StableHLO to SavedModel, also include the original var (#5711)

haozha111 committed 2 years ago

Verified e7af313d

Remove GetTensorsFused since we don;t have opbyop anymore (#5718)

JackCaoG committed 2 years ago

Verified 5dfcec90

fix typo in TROUBLESHOOTING (#5716)

JackCaoG committed 2 years ago

Verified 4fa727b4

Run decomp before processing (#5713)

qihqi committed 2 years ago

Verified b1123dbb

Add support for `_unsafe_index`. (#5707)

ysiraichi committed 2 years ago

Verified bccbb5aa

Add multi-host GPU support (#5657)

vanbasten23 committed 2 years ago

Verified 6ea99474

Update Troubleshooting with some sanity check example (#5705)

JackCaoG committed 2 years ago

Verified 8f45caef

Older