strisunshinewentingwang/pytorch

Pull Requests Commits

[FSDP()][7/N] Refactor most of ctor (#87921)

awgu committed 1 year ago

d89cf2fd

[FSDP()][3/N] Refactor public APIs (#87917)

awgu committed 1 year ago

9d9267c6

Fix: prefer .is_none() over .is(py::none()) for pybind11 (#88051)

Skylion007 committed 1 year ago

59fe272c

make autocast cache global instead of thread-local (#86492)

vkuzo committed 1 year ago

75dbe379

[FSDP] Enable `use_orig_params=True` test (#88034)

awgu committed 1 year ago

34f523b2

[Vulkan] Add Vulkan Rewrite to Transfer Inputs and Outputs to Vulkan and CPU Backends Respectively (#87432)

salilsdesai committed 1 year ago

df1cc0ef

[Vulkan] Add support for Optimization Blocklist to Vulkan Rewrite (#87431)

salilsdesai committed 1 year ago

bc686251

.gitignore log files (#88085)

ezyang committed 1 year ago

f717986f

Make IValue::unsafeToTensorImpl a little less unsafe. (#88043)

ezyang committed 1 year ago

8ea19c80

Simplify magic method definition code. (#88017)

ezyang committed 1 year ago

e238752e

Get the magic method try reverse protocol correct (#88030)

ezyang committed 1 year ago

2a47b107

Fix all references to torchdynamo from the merge (#87731)

Chillee committed 1 year ago

12dd8773

Add fake tensor files to ciflow/inductor (#88052)

ezyang committed 1 year ago

496acb66

[test_nn] split convolution tests from test_nn (#87474)

kshitij12345 committed 1 year ago

6735bf21

fix github bug issue 87552 (#88059)

jingxu10 committed 1 year ago

46ce9271

Use scaled_dot_product_attention within attention.cpp (#87312)

drisspg committed 1 year ago

e24ce484

Add sequence number support for UCC (#85047)

Fuzzkatt committed 1 year ago

d13f1e6a

[ONNX] Fix get wrong summary of the docstring in `torch.onnx._deprecation.deprecated` (#87194)

HAOCHENYE committed 1 year ago

9642a7c2

[dynamo][dashboard] minor fixes for a clean Dashboard (#88056)

anijain2305 committed 1 year ago

d67b2ede

Even "nvcc not found" should be commented out (#87959)

brad-mengchi committed 1 year ago

9109ecf9

[dynamo][benchmarks] use fresh inductor cache and raise batch size wherever possible (#88044)

anijain2305 committed 1 year ago

1b575782

[BE] Do not package caffe2 in wheel (#87986)

malfet committed 1 year ago

e7b854fa

[vision hash update] update the pinned vision hash (#87948)

pytorchmergebot committed 1 year ago

65e77195

[BE] Do not assign string literal to `char *` (#87949)

malfet committed 1 year ago

621158cd

[Inductor] Enable Inductor unspec inputs test for different dtypes (#87809)

yanboliang committed 1 year ago

59001d05

Revert "Unify meta tensor and fake tensor converter conversion (#87943)"

pytorchmergebot committed 1 year ago

bc64999b

torchdynamo and xla integration (#87741)

shunting314 committed 1 year ago

e4a8661a

Use TORCH_CHECK instead of inappropriate CUDA_KERNEL_ASSERT (#87714)

r-barnes committed 1 year ago

6cd25eb6

[BE] Upload GHA artifacts to S3 (#87827)

huydhn committed 1 year ago

384b84d6

Add composable activation checkpointing (#87664)

mrshenli committed 1 year ago

d9b6e41d

Newer Older