pytorch/xla

Pull Requests Commits

[torch_ xla2] LLama3 example - fixed sharding configuration on LLama3 example (#7973)

zmelumian committed 1 year ago

Verified ef6b3703

[torch_xla2] Fix the test of `cumsum` (#7965)

guyao committed 1 year ago

Verified 12e59584

Update Neuron initializations (#7952)

jeffhataws committed 1 year ago

Verified 901c3a32

grid_sampler_3d, gaussian_nll_loss, dropout (#7969)

qihqi committed 1 year ago

Verified 900296aa

Remove `TPU_LIBRARY_PATH` from legacy TPU CI (#7968)

will-cromar committed 1 year ago

Verified 0528c9d7

Increment version in upstream CI `docker push` command (#7967)

will-cromar committed 1 year ago

Verified c201ec11

Revert "Allow MpDeviceLoader to shard dictionaries of tensors" (#7964)

JackCaoG committed 1 year ago

Verified 59570c77

Unpin `scipy` in upstream CI image (#7963)

will-cromar committed 1 year ago

Verified 282bf932

Run ansible release stage after local builds are installed (#7955)

will-cromar committed 1 year ago

Verified 8bebeddd

Remove numpy pin for upstream image (#7954)

will-cromar committed 1 year ago

Verified 92f8cfdf

Op info test for logaddexp, logaddexp2 and logcumsumexp (#7506) (#7957)

yenkwang committed 1 year ago

Verified 1ae32bf9

Update upstream docker image to Python 3.10 (#7953)

will-cromar committed 1 year ago

Verified 1a417b4b

Op info test for special.modified_bessel_i1 .. special.polygamma (#7928)

ayshiff committed 1 year ago

Verified e7653b85

Allow MpDeviceLoader to shard dictionaries of tensors (#7912)

bhavya01 committed 1 year ago

Verified 5acecc33

`DynamicShapeDetector` with trie implementation. (#7918)

ysiraichi committed 1 year ago

Verified 400bd91b

support dist.reduce_scatter_tensor (#7950)

zpcore committed 1 year ago

Verified b1fbb749

Skip dynamo collective tests on TPU v2 CI (#7949)

will-cromar committed 1 year ago

Verified f45775a5

bring XLA_USE_BF16 and XLA_DOWNCAST_BF16 back for 2.5 release (#7945)

JackCaoG committed 1 year ago

Verified d224c7a3

Include Terraform lock for release triggers (#7948)

will-cromar committed 1 year ago

Verified 2540ae78

Ignore torch_xla2 on legacy TPU CI (#7947)

will-cromar committed 1 year ago

Verified 3a64bbc8

isclose (#7941)

ShitongWang committed 1 year ago

Verified 2e34aea3

fix dynamo inplace copy (#7933)

zpcore committed 1 year ago

Verified 989ac69c

Add `tpu-info` as a dependency of `torch_xla[tpu]` (#7938)

will-cromar committed 1 year ago

Verified d69f3a54

Fix jaten for softmax(nan/inf). (#7944)

tqfan28 committed 1 year ago

Verified 9e3b4484

Op info test for expm1 .. fill (#7940)

qihqi committed 1 year ago

Verified 13affb9d

[torch_xla2] Ops fixes https://github.com/pytorch/xla/issues/7512 (#7936)

ShitongWang committed 1 year ago

Verified 87fe71ab

Double loader_prefetch_size and device_prefetch_size (#7934)

tengyifei committed 1 year ago

Verified a0b20edb

Support gmm and tgmm trace_pallas caching (#7921)

JackCaoG committed 1 year ago

Verified 8955571f

[benchmarks] Fix processing of PyTorch profiler events. (#7930)

ysiraichi committed 1 year ago

Verified 1b4b8281

[benchmarks] Add IR dump option. (#7927)

ysiraichi committed 1 year ago

Verified beeb40b1

Newer Older