strisunshinewentingwang/pytorch

Pull Requests Commits

[NVFuser] always use fallback if fusion fails

davidberard98 committed 2 years ago

da984c50

Distribute torchgen as part of PyTorch package

ezyang committed 2 years ago

5109d81f

Revert "[nnc] Strides to Tensor (#72962)"

zengk95 committed 2 years ago

1d555181

[flatbuffer] Bugfix: some class dont have __getstate__ (#76197)

qihqi committed 2 years ago

c55b425d

[FSDP] Support initialization of modules on meta device

rohan-varma committed 2 years ago

b4b1b0d5

Make sure requires_grad is propagated for all backend

albanD committed 2 years ago

e4d5801e

Re-land D35572928: [kineto] implement kineto client interface (#76239)

chaekit committed 2 years ago

920876d6

[Vulkan] VK Timestamp Queries for op profiling (#75829)

beback4u committed 2 years ago

4f3dd80c

move Bazel version header generation to shared build structure (#75332)

mikey dagitses committed 2 years ago

f4200600

[cmake] Use list(APPEND instead of string(APPEND for vulkan codegen args

smspillaz committed 2 years ago

111b2bf9

delete ${GEN_VULKAN_FLAGS}

MambaWong committed 2 years ago

bc9bba9b

add comment for lerp cuda implementation

jjsjann123 committed 2 years ago

7a27accb

define the caffe2_serialize target in Bazel (#75942)

mikey dagitses committed 2 years ago

d78dd825

[JIT] Add Context Manager to force strict fusion

Elias Ellison committed 2 years ago

0d7be81c

[EASY] [JIT]Cleaning up the device analysis temp test code

Gamrix committed 2 years ago

fdaf393e

Revert "[PyTorch] Add native fast path for transformer encoder inference"

bigfootjon committed 2 years ago

2387efd3

remove bucket_size_limit property from bucket struct

zhaojuanmao committed 2 years ago

1c35f37c

[composite compliance] add test for fwd AD

kshitij12345 committed 2 years ago

c1ced8ff

ROCm: Enable test_distributed_spawn

jaglinux committed 2 years ago

81586a6a

ROCm: unskip c10 gloo tests

jaglinux committed 2 years ago

60e2ee39

[DataPipe] Add funtional API docstring and fix typo in test

ejguan committed 2 years ago

0ff05b1e

Make -h work with run_test.py

albanD committed 2 years ago

3d7abc0e

Improve more the error message with explicit recommendation

albanD committed 2 years ago

438cc79f

[shard] Sharder and ShardingPlan prototype (#73873)

wanchaol committed 2 years ago

78ea86a4

[FSDP] Fix exec order validation (static variable issue)

awgu committed 2 years ago

f7e4b77a

Initial prims, references, and test architecture for them (#75095)

Mike Ruberry committed 2 years ago

28c3e0f7

[PyTorch] Add native fast path for transformer encoder inference

swolchok committed 2 years ago

b369b89f

Rename tools/codegen to torchgen (#76275)

ezyang committed 2 years ago

36420b5e

[ONNX] Support restricted quantized range for activation.

BowenBao committed 2 years ago

8d31706b

[ONNX] Support per channel quantization

BowenBao committed 2 years ago

cada2cd3

Newer Older