strisunshinewentingwang/pytorch

Pull Requests Commits

Modernize fix deprecated header (#61736)

r-barnes committed 3 years ago

59a5312c

Modernize some loops in torch (#61737)

r-barnes committed 3 years ago

5a04bd87

[Docs] Bundle of errata and small corrections / improvements for torch.linalg docs (#61578)

lezcano committed 3 years ago

65616184

comment out failed tests

tktrungna committed 3 years ago

1d265d11

Merge branch 'master' into build-whl-for-testing

tktrungna committed 3 years ago

a1652bb7

bitwise_and: Port to structured (#60813)

Freey0 committed 3 years ago

a0c9d70f

bitwise_xor: Port to structured (#60812)

Freey0 committed 3 years ago

875d63ed

bitwise_or: Port to strucutred (#60811)

Freey0 committed 3 years ago

ce8aeefb

Add thread local state guards in autograd engine hooks. (#60067)

chengjunlu committed 3 years ago

f59ac5ab

Implement IMethod::getArgumentNames() (#61856)

Jiewen Tan committed 3 years ago

641f6ef8

[bc-breaking] Dispatch index_put with boolean mask argument to masked_fill (#61612)

eqy committed 3 years ago

42d6543c

Factor vector intrinsics out of SumKernel.cpp (#61483)

peterbell10 committed 3 years ago

018dc419

Use cascade-summation to improve nansum accuracy (#61082)

peterbell10 committed 3 years ago

c44d9d9f

logit_backward: Port to structured (#60817)

Freey0 committed 3 years ago

bf1c9aaa

tanh_backward: Port to structured (#60816)

Freey0 committed 3 years ago

b8686b42

sigmoid_backward: Port to structured (#60815)

Freey0 committed 3 years ago

8c42d7ad

xlogy: Port to structured (#60814)

Freey0 committed 3 years ago

11cc1793

Makes a streaming backward test try gradient stealing more directly (#60065)

mcarilli committed 3 years ago

9fb6b40f

Support 3 argument variant of the getattr() call where the third arg is the default return value (#61599)

Jerry Cai committed 3 years ago

873cc7a4

[CUDA graphs] Make sure graph mempool cudaMalloc_count decrement pairs with cudaFree for all allocations (#61567)

mcarilli committed 3 years ago

ffd2e602

Port other comparison ops: `ne`, `lt`, `gt`, `le`, `ge` to structured kernels. (#60942)

ysiraichi committed 3 years ago

208d06ca

Port `eq` kernel to structured kernels. (#60177)

ysiraichi committed 3 years ago

97327137

[vulkan] Adaptive local work group size (#61170)

SS-JIA committed 3 years ago

64ac4288

[vulkan] Calculate a 4x4 output tile for each invocation in conv2d_pw (#60760)

SS-JIA committed 3 years ago

f324421d

[vulkan] Convolution op cleanup (#60759)

SS-JIA committed 3 years ago

a1b5025e

[vulkan] Reduce submission rate to save CPU cycles (#60758)

SS-JIA committed 3 years ago

cacab7e9

[package] merge test_torchscript into test_package_script (#61807)

suo committed 3 years ago

554038c2

ban PyTorchStreamWriter from writing the same file twice (#61805)

suo committed 3 years ago

f02cfcc8

[package] fix storage serialization collision (#61806)

suo committed 3 years ago

04043d68

add BFloat16 support for topk on CPU (#59547)

jiayisunx committed 3 years ago

c30048fc

Newer Older