strisunshinewentingwang/pytorch

Pull Requests Commits

Allow Process Group to support multiple backends (#88330) (#90997)

H-Huang committed 2 years ago

7a0f29b7

[dynamo] Refactor how autocast parameters are binded (#90953)

desertfire committed 2 years ago

93ac8c4a

Add macro C10_AS_INTARRAYREF_SLOW (#90675)

ezyang committed 2 years ago

Verified 4fa8d774

Revert "[cuDNN][cuDNN V8 API] (re-re-open) cuDNN V8 API on by default (#89022)"

pytorchmergebot committed 2 years ago

ba7aeac3

Fix non-existing parameters in docstrings in torch/ao (#90875)

kit1980 committed 2 years ago

4438b019

ModuleInfo-based tests for AOTAutograd (#90980)

jbschlosser committed 2 years ago

ee247586

LSTM SymInt-aware changes & meta registration (cuDNN) (#90944)

jbschlosser committed 2 years ago

32262096

Introduce causal mask (#90508)

Michael Gschwind committed 2 years ago

512ec181

Don't recompute var in bn decomp (#90984)

Natalia Gimelshein committed 2 years ago

e689c509

functorch: non-contig samples for test_grad (#90990)

kshitij12345 committed 2 years ago

e4de6ed6

[FSDP][3/N] Move `fsdp_modules(root_only=True)` -> `_get_fsdp_root_states()` (#90862)

awgu committed 2 years ago

5ea418bf

Revert "[Quant] onednn backend switch to ideep new api without affacting performance (#90354)"

pytorchmergebot committed 2 years ago

67ef88af

aot_autograd: add assert for functional-only graph (#88816)

bdhirsh committed 2 years ago

7a683eae

[GHA][BE] Update to newer checkout action (#90969)

malfet committed 2 years ago

c83ff1ea

[quantized] [executorch] typo (#89960)

JacobSzwejbka committed 2 years ago

bd94ee66

Include dispatch key in wrapper symbol name (#90674)

ezyang committed 2 years ago

68805b56

Revert "[reland2][dynamo] Revert "Revert "[reland][dynamo] use optimizers correctly in benchmar… (#90956)"

pytorchmergebot committed 2 years ago

6bc6fb21

[FSDP] Introduce "fully sharded module"; remove comm. module (#90933)

awgu committed 2 years ago

8cd1808d

LSTM SymInt-aware changes & meta registration (non-cuDNN CUDA) (#90701)

jbschlosser committed 2 years ago

b0cda0b3

generate device context managers in inductor code (#90934)

Natalia Gimelshein committed 2 years ago

a10b3ce8

tools: Update clang-tidy hash (#91008)

seemethere committed 2 years ago

9d8fa78d

Ensure sorted indices from the CSR->BSR conversion (#90918)

pearu committed 2 years ago

01e7f462

[ONNX] Auto test based on OpInfo (#86182)

justinchuby committed 2 years ago

634555d9

[reland2][dynamo] Revert "Revert "[reland][dynamo] use optimizers correctly in benchmar… (#90956)

mlazos committed 2 years ago

8bc38ae4

Sparse compressed mm: fix for orthogonal inputs (#90917)

nikitaved committed 2 years ago

c2c14f95

Sparse compressed mm: fix for empty inputs (#90763)

nikitaved committed 2 years ago

4dd3de23

[FSDP][2/N] Move `fsdp_modules(root_only=False)` -> `_get_fsdp_states()` (#90861)

awgu committed 2 years ago

3e44fcee

[FSDP][Easy] Rename `entry` -> `fsdp_module` to be more descriptive (#90864)

awgu committed 2 years ago

673c25d4

[FSDP][1/N] Add `_get_fsdp_states()` (#90860)

awgu committed 2 years ago

95ee5fec

[Inductor] actually check `replacements` in `AutogradMonkeypatch` (#90901)

nkaretnikov committed 2 years ago

06533a2e

Newer Older