pytorch/xla

Pull Requests Commits

Fix incorrect xla type for view_as_complex

JackCaoG committed 1 year ago

bef00774

Support the case where query_len>context_len in the multi-queries paged attention. (#8356)

vanbasten23 committed 1 year ago

Verified 102cd489

Expose lazy all number special scalars (#8361)

rpsilva-aws committed 1 year ago

Verified f610f704

[torch_xla2] Enable nn.functional.cosine_embedding_loss (#8368)

ManfeiBai committed 1 year ago

Verified d4c90c91

Fix op info test for svd_lowrank (#8376)

dvhg committed 1 year ago

Verified 7220aee4

Revert S64 and U64 type change for Neuron device due to hardcoded dtype checks in torch-xla (#8375)

aws-nm9 committed 1 year ago

Verified bc227f7f

Remove unnecessary prints in PagedAttention (#8374)

WoosukKwon committed 1 year ago

Verified baa14f36

Fix XLA type of the view_as_real (#8370)

JackCaoG committed 1 year ago

Verified 071ddfa6

Improve some typing annotations (#8369)

tengyifei committed 1 year ago

Verified c4f27718

Add tengyifei to docs code owners (#8365)

tengyifei committed 1 year ago

Verified 5a4641b4

Add mikegre-google as docs owner (#8364)

tengyifei committed 1 year ago

Verified 09f36bee

XLA downcast for S64 and U64 for Neuron (#8347)

aws-nm9 committed 1 year ago

Verified 03f07e2a

Add aten::lift_fresh to the exception list that converts input to torch tensor (#8360)

barney-s committed 1 year ago

Verified 6454b42f

Add inplace ops for sigmoid and tanh (#8358)

barney-s committed 1 year ago

Verified 77abab25

Revive "Modularize zero step function and make it customizable" #7233 (#8332)

jeffhataws committed 1 year ago

Verified 7d6a4f25

Make XLA tests export new IR compatible (#8354)

tugsbayasgalan committed 1 year ago

Verified 81c4caa7

Correct the cxx11_abi condition (#8357)

tengyifei committed 1 year ago

Verified ca0a244a

Remove unnecessary functions and print statements (#7495) (#8352)

simonteozw committed 1 year ago

Verified 57cce803

Fix torch.linalg.matrix_power (#7495) (#8351)

simonteozw committed 1 year ago

Verified 7dda4373

Add functions to emit custom call to place a buffer to host and device. (#8350)

qihqi committed 1 year ago

Verified a19a996f

[Neuron] Fix the XLADevice Neuron mappings for SPMD downcasts (#8335)

rpsilva-aws committed 1 year ago

Verified 7c7ad4e7

Extend paged attention to support query_len>1 (#8328)

vanbasten23 committed 1 year ago

Verified 1bac0626

Remove torch_pin that is accidentally introduced here: (#8344)

qihqi committed 1 year ago

Verified efee3bc5

Add support for native_layer_norm (#8031)

matinehAkhlaghinia committed 1 year ago

Verified 57dbfd7e

update runtime error message for minibatch (#8243)

JackCaoG committed 1 year ago

Verified 3efe1ebf

DLPack: add test using PyTorch DLPack functions. (#8294)

ysiraichi committed 1 year ago

Verified dc20b2d9

Fixes issue #7774 (#8322)

mikegre-google committed 1 year ago

Verified 81774476

Adds torch.nn.functional.cosine_similarity Jax implementation (#8305)

jimlinntu committed 1 year ago

Verified c9e02e9d

[2.5.1 release] Update artifacts.auto.tfvars (#8339)

ManfeiBai committed 1 year ago

Verified 89e47b33

[2.5.1 release] Update artifacts.auto.tfvars for 2.5.1 release (#8336)

ManfeiBai committed 1 year ago

Verified d9fb89d6

Older