strisunshinewentingwang/pytorch

Pull Requests Commits

Don't store device indices separately on FutureNCCL (#48501)

lw committed 3 years ago

9fe3ac36

Add multi-GPU support to FutureNCCL (#48500)

lw committed 3 years ago

e294c2d8

Fix FutureNCCL not recording dataptrs with caching alloc in wait() (#48563)

lw committed 3 years ago

91ad3ed8

Fix FutureNCCL's completed() disagreeing with wait() (#48503)

lw committed 3 years ago

003c30ba

Record CUDA events for "follow-up" FutureNCCL inside markCompleted (#48499)

lw committed 3 years ago

b91b0872

Use fresh stream from pool for each FutureNCCL callback (#48498)

lw committed 3 years ago

6157f8ae

Make FutureNCCL record events in current stream (#48497)

lw committed 3 years ago

8fb52e7f

Have FutureNCCL record streams w/ allocator in addCallback (#48496)

lw committed 3 years ago

e4267eb4

Add some safeguards to FutureNCCL (#48562)

lw committed 3 years ago

868a1a48

Remove NCCL dependency from PythonFutureWrapper (#48495)

lw committed 3 years ago

b7f5aa98

Avoid using FutureNCCL before it's ready (#48561)

lw committed 3 years ago

7f7f0fa3

[numpy] `torch.exp{2, m1}`: promote integer inputs to float (#48926)

kshitij12345 committed 3 years ago

eb9516ea

Port `eig` CPU from TH to ATen (#43215)

kurtamohler committed 3 years ago

27f7d1c2

[PyTorch Mobile] Preserve bundled input related methods when calling optimize_for_mobile

bearzx committed 3 years ago

95233870

op to gen quant params from min-max thresholds

Xiaohan Wei committed 3 years ago

9417e927

[NNC] Dead Store Elimination (#49030)

nickgg committed 3 years ago

c5bc6b40

Revert D25416620: [pytorch][PR] Add version_info tuple

ezyang committed 3 years ago

7a2abbd8

[PyTorch] Avoid storage refcount bump in copy_tensor_metadata (#48877)

swolchok committed 3 years ago

3123f878

Add version_info tuple (#48414)

jonashaag committed 3 years ago

e69c2f85

Add type annotations to conv-relu (#47680)

guilhermeleobas committed 3 years ago

5375a479

[PyTorch Mobile] Add continuous build config for xplat/caffe2

dhruvbird committed 3 years ago

e9ef1fe3

Class-based structured kernels, with migration of add to framework (#48718)

ezyang committed 3 years ago

16b8e6ab

adding profile_ivalue (#47666)

jjsjann123 committed 3 years ago

a6fa3b26

[collect_env] Acquire windows encoding using OEMCP (#49020)

peterjc123 committed 3 years ago

f431e47a

Review memory overlap checks for advanced indexing operations (#48651)

peterbell10 committed 3 years ago

5765bbd7

[PyTorch] Remove aten::native::empty usage in TensorIndexing (#49074)

iseeyuan committed 3 years ago

dfa38087

migrating some straggler pytorch ops in fbcode to the new registration API (#48954)

bdhirsh committed 3 years ago

c7cc8a48

Pass shape hints for AOT case (#48989)

khabinov committed 3 years ago

67d12c95

Revert D25325039: Check CUDA kernel launches (/fbcode/caffe2/)

supriyar committed 3 years ago

bfa95f90

Revert D25003113: make validate debug-only in Device copy ctr

supriyar committed 3 years ago

7a4a2df2

Newer Older