pytorch/xla

Pull Requests Commits

update dynamo debug message (#6629)

JackCaoG committed 1 year ago

Verified 8be7f269

Add test for Gemma model (#6631)

qihqi committed 1 year ago

Verified 74ba968e

AliasWithBufferDonorContext wtih default value (#6628)

JackCaoG committed 1 year ago

Verified c99ed84c

Add UInt16/32/64 support (#6626)

lsy323 committed 1 year ago

Verified 0a2e5b5a

Use BERT for the HF Transformers example (#6584)

will-cromar committed 1 year ago

Verified 3c8c54e4

BufferDonor based aliasing enhancement (#6619)

JackCaoG committed 1 year ago

Verified 472d90d0

Change to correct directory before running v2 TPU CI (#6622)

will-cromar committed 1 year ago

Verified db64a454

Automatically detect CUDA if it isn't set and default to upstream torch.cuda number of devices (#6605)

changm committed 1 year ago

Verified cb4983e9

[benchmarks] Run some models with smaller batch sizes. (#6542)

ysiraichi committed 1 year ago

Verified cd47390f

Add API to donate input buffer for dynamo execution (#6587)

JackCaoG committed 1 year ago

Verified 3e2a23ca

Shorten ResNet tests and add more test coverage. (#6606)

vanbasten23 committed 1 year ago

Verified 65f54651

Move TPU CI into shared script (#6602)

will-cromar committed 1 year ago

Verified bc3615c8

Update nightly to 2.3 (#6617)

JackCaoG committed 1 year ago

Verified 9b81acc0

nightly.sh: build with local compute capability (#6608)

cota committed 1 year ago

Verified 4327d24a

nightly.sh: fix assignment of timestamp default (#6601)

cota committed 1 year ago

Verified a9a788d5

Add ResNet18 per channel quant test, fix assertion in quantize func and name mangling for tf saved model (#6592)

lsy323 committed 1 year ago

Verified 1654ae7f

Fix instruction for getting HLO/IR dump (#6599)

vanbasten23 committed 1 year ago

Verified b219da81

Add topological sort after composite grouping pass (#6607)

lsy323 committed 1 year ago

Verified dc08cb68

[benchmarks] Set autocast kwargs only if AMP. (#6612)

ysiraichi committed 1 year ago

Verified f65bae4d

Conditional run tests that have TF dependency (#6573)

lsy323 committed 1 year ago

Verified 8217c6b5

Make sure torch_xla2 works with torch stable 2.2 (#6593)

qihqi committed 1 year ago

Verified a659d970

[benchmarks] Fix YAML file name. (#6597)

ysiraichi committed 1 year ago

Verified 1a2ea9de

[Core Aten Ops] Lower reflection_pad1d, reflection_pad1d_backward, reflection_pad3d and reflection_pad3d_backward (#6588)

ManfeiBai committed 1 year ago

Verified 13e8647c

Allow Inductor benchmarks for CPU backends. (#6555)

changm committed 1 year ago

Verified 263da4ee

Update OpenXLA pin to fix GPU build (#6569)

cota committed 1 year ago

Verified 2bbc9a49

[Fori_loop|While_loop] Placeholder lower torch.while_loop with python dispatch for simple addition test case (#6532)

ManfeiBai committed 1 year ago

Verified 5405c4d4

Enable graph input matcher test (#6591)

JackCaoG committed 1 year ago

Verified 69d453ea

Add extra APIs. (#6586)

qihqi committed 1 year ago

Verified 5145ea1c

Add XLA specific XPlane profiling. (#6552)

golechwierowicz committed 1 year ago

Verified adabb59a

Fix some more core aten ops (#6580)

wonjoo-wj committed 1 year ago

Verified 2cb31440

Newer Older