strisunshinewentingwang/pytorch

Pull Requests Commits

[vision hash update] update the pinned vision hash (#95340)

pytorchmergebot committed 1 year ago

c594a32f

[SDPA] Fix bug in parsing scaled_dot_product_attention arguments (#95311)

drisspg committed 1 year ago

29c235e5

use 4 warps for small block config in mm (#95339)

ngimel committed 1 year ago

8e391c73

[inductor] enable `test_nll_loss_forward_dynamic_shapes` (#95176)

nkaretnikov committed 1 year ago

ba8ff4be

Fix disbale typos (#95322)

kit1980 committed 1 year ago

f98733e9

Bugfix nested mem_efficient path in SDPA when E_qk != E_v (#95330)

mikaylagawarecki committed 1 year ago

586ac98c

[FSDP][Docs] Re-add why reg. post-bwd hook on 1st forward (#95326)

awgu committed 1 year ago

78175cee

Avoid FPE when running batch norm with zero batch size. (#95324)

ezyang committed 1 year ago

f247129f

coreml_delegate - Add input shape in error when throwing from predicting (#95249)

kylesyoon committed 1 year ago

a257486b

[pt2][inductor] switch dinfo representation (#95302)

nmacchioni committed 1 year ago

3ebab9ae

Preserve meta["val"] on export (#95314)

ezyang committed 1 year ago

ca7eb1ba

Second part of splitting #91254 in two (#92749)

Fabio Rocha committed 1 year ago

f6f413c6

[BE] Simplify `Source.is_nn_module`; add some types (#95292)

awgu committed 1 year ago

cbac56e2

Make fx.Transformer.get_attr call tracer to preserve node.meta (#95245)

ydwu4 committed 1 year ago

674ef1f9

Update isend/irecv warning messages for nccl (#95236)

H-Huang committed 1 year ago

c0fa0669

Implement sparse semantics support in gradcheck (#94714)

pearu committed 1 year ago

7ac511c2

[MPS] Remove mps specialized path in BCE backward (#95220)

qqaatw committed 1 year ago

b6a1c238

[MPS] Add xlogy op (#95213)

qqaatw committed 1 year ago

69c76ff0

[DCP][nit] Rename variables + minor documentation fix for optimizer.py (#95264)

wz337 committed 1 year ago

5fa93788

Reland "Introduce constrain_range; remove old expr_subs (#95063)" (#95209)

ezyang committed 1 year ago

3758559a

Fix formatting for merge failed message (#95234)

ZainRizvi committed 1 year ago

d6a8d397

[MPS] Fix Float16 issue with Reduction ops for macOS 12 (#94952)

razarmehr committed 1 year ago

d88d4145

[MPS] Convolution cleanup; remove unnecessary contiguous calls (#95078)

DenisVieriu97 committed 1 year ago

5e47571a

[MPS] Handle broadcasting by expanding src tensor in Copy.mm (#95272)

kulinseth committed 1 year ago

02a6d433

During export, generate Python TENSOR_MATCH guards (#94970)

voznesenskym committed 1 year ago

5a8092f0

[MPS] Cast int64 to int32 for reduction ops (#95231)

DenisVieriu97 committed 1 year ago

8475af77

Revert "During export, generate Python TENSOR_MATCH guards (#94970)"

pytorchmergebot committed 1 year ago

6ae60b19

[pt2][inductor] search caches by default (#95134)

nmacchioni committed 1 year ago

5f24b2b1

Add dynamo bench arg --per_process_memory_fraction (#95260)

wconstab committed 1 year ago

8de4238a

Support registering op returning symint in python (#95240)

shunting314 committed 1 year ago

a4b02a15

Newer Older