strisunshinewentingwang/pytorch

Pull Requests Commits

add pytorchbot to merge rules for triton pin, add ciflow/inductor automatically on pin changes

clee2000 committed 1 year ago

c21200e0

[vmap] Fix searchsorted batch rule (#99698)

qqaatw committed 1 year ago

a67fa845

Do not use `--extra-index-url` in testing wheels (#100183)

malfet committed 1 year ago

991b1c02

[quant][pt2e] remove dropout from fx quant

mcr229 committed 1 year ago

Verified 151d76cc

Optimize periodic jobs (#100182)

malfet committed 1 year ago

089b085c

[SPMD][Easy] Add time counter in graph_optimization_pass (#99969)

fegin committed 1 year ago

01de8ee8

[DDP] Perform input casting in pre forward (#100131)

rohan-varma committed 1 year ago

87db02ea

[Experimental] Remove store barrier after PG init (#99937)

kwen2501 committed 1 year ago

ae0eb234

[export] Port over const prop pass (#100102)

angelayi committed 1 year ago

7bece142

[PTD][Checkpoint] Upstream fsspec storage read/write to PT (#98387)

wz337 committed 1 year ago

fad2f6ed

[SPMD] Add embedding dense backward prop rule for postional embedding (#100038)

fegin committed 1 year ago

b94a0ba5

Remove CUDA 11.6 note from complex docs (#100118)

ptrblck committed 1 year ago

8fe91d16

Add private _export API (#99992)

tugsbayasgalan committed 1 year ago

02f059c2

[dynamo][numpy] Handle return value being numpy ndarray (#99560)

larryliu0820 committed 1 year ago

f5853342

[dynamo][numpy] Add NumpyTensorVariable to translate ndarray attribute calls to tensor attributes (#95849)

larryliu0820 committed 1 year ago

687afeb6

[Dynamo] Add unit test for explicitly calling __call__ (#100146)

yanboliang committed 1 year ago

d855b6ae

Fix cat forward-AD tests (#99596)

qqaatw committed 1 year ago

cb569dbc

[inductor] Fix argmin/max with duplicate values (#99920)

peterbell10 committed 1 year ago

659dcc5e

[inductor] Fix nan-handling of max and min reductions (#99881)

peterbell10 committed 1 year ago

f9c3fcd1

[inductor] Create triton_helpers module for helper functions (#99880)

peterbell10 committed 1 year ago

ed2eb13d

[c10d] Scalable PG initiation. (#99931)

Rodrigo Kumpera committed 1 year ago

ad21890f

sparse.sum backward: short circuit on zero/empty grad (#98838)

nikitaved committed 1 year ago

2eab5abb

Add support for serializing real tensor data in after aot minifier (#99834)

ezyang committed 1 year ago

67e0913d

relu/threshold backward for sparse: enable 0-nnz grads (#98935)

nikitaved committed 1 year ago

5cfaea15

Change caffe2 branch links to main (#100129)

kit1980 committed 1 year ago

c2402a92

Include all mkl/mkldnn related test files to CPU ATen backend (#99592)

jgong5 committed 1 year ago

77a37a54

Basic dynamo support for traceable collectives (#94440)

wconstab committed 1 year ago

100a25d0

[CUDA] Switch to `at::empty` in `max_pool3d_with_indices_backward_cuda` (#100138)

eqy committed 1 year ago

925a3788

Making fsdp device-agnostic for custom-backend which implement cuda-semantics (#99024)

medivh-xp committed 1 year ago

859e82a7

[inductor] fix _print_Pow given reciprocal of dynamic dim with float exponent (#100090)

jgong5 committed 1 year ago

4456e932

Older