strisunshinewentingwang/pytorch

Pull Requests Commits

Added possibility to index scalars by bool masks (#21030)

izdeby committed 5 years ago

00c15849

Improve test_proper_exit error printing (#20166)

ssnl committed 5 years ago

1d4685c2

ctc_loss: fix backward when 2d target tensor is larger than max_target_length (#20971)

t-vi committed 5 years ago

aa42742d

DilatedMaxPool2d: small cleanup

skrah committed 5 years ago

55f5eb3c

Port dilated_max_pool3d() to ATen

skrah committed 5 years ago

f8565121

Revert D15499749: [pytorch][PR] Add `Tensor.T` attribute to reverse dimensions

ezyang committed 5 years ago

0544a491

Remove THSTensor and SparseTensorRef (#20877)

Roy Li committed 5 years ago

3038cf8e

Fix __irshift__ dispatch (#21047)

Roy Li committed 5 years ago

9faa409b

Remove extraneous TensorId checks in as_strided (#21045)

Roy Li committed 5 years ago

8dda19b7

Fix tracing bugs where using `1 - x` in C++ would cause the size of 1 to get hardcoded. (#20932)

zdevito committed 5 years ago

d76546a4

Make build with makefiles less noisy (#21053)

bddppq committed 5 years ago

5c53aa48

Fix get_gpu_memory_info in non-cuda builds (#21054)

bddppq committed 5 years ago

9b147961

Deprecate variadic inputs of checkpoint_sequential (#21006)

sublee committed 5 years ago

ffdce790

Allow nondet_tol for nondeterminism in gradcheck and gradgradcheck (#20980)

t-vi committed 5 years ago

d23d04f1

Fix typo in CyclicLR docs (#21021)

njdalton committed 5 years ago

d190450a

add simple memory analyzer and log warning if GPU underutilized (#21024)

Aapo Kyrola committed 5 years ago

f1fe4b11

Fix warning in register_c10_ops by making index unsigned (#20964)

Chillee committed 5 years ago

1bed5f39

Enable batched QR decomposition and add a `some` option (#20689)

vishwakftw committed 5 years ago

f6ec4648

Use constants from math.h for gelu op (#20974)

xiaomengy committed 5 years ago

c1048182

tracing for intra_op_parallel (#20603)

jspark1105 committed 5 years ago

0290897b

Add an option to stop the build process once cmake terminates. (#21034)

xuhdev committed 5 years ago

9a989ec4

Add `Tensor.T` attribute to reverse dimensions (#20598)

Brennan Vincent committed 5 years ago

9294de8c

Renaming the relu kernel and adding hypothesis tests (#20647)

Zafar Takhirov committed 5 years ago

2791a449

Added engine information to the profiling result. (#20493)

kimishpatel committed 5 years ago

d6d192e0

Enable operator profiling via command line (#20173)

kimishpatel committed 5 years ago

7afa7500

Fixed gcd to use 64 bit integers (#21041)

Chillee committed 5 years ago

2ba608b4

Make ScriptModule.training an attribute instead of a parameter (#19587)

David Riazati committed 5 years ago

28079c39

Better error message in __get_state__ to let a user know that ScriptModules can't be deep-copied atm (#20885)

Krovatkin committed 5 years ago

18809f7b

Some minor fixes for the changes in #20945 (#21008)

peterjc123 committed 5 years ago

07c4e45c

refactor register_prim_ops (#21001)

wanchaol committed 5 years ago

0885dd28

Newer Older