strisunshinewentingwang/pytorch

Pull Requests Commits

Emit std::string instead of c10::string_view for Lazy IR class

desertfire committed 2 years ago

46df3997

opinfo tests (#73979)

Krovatkin committed 2 years ago

Verified 81962efd

[LT] run_tests.sh for python3.9 (#73887)

davidberard98 committed 2 years ago

Verified 4868b47d

Implement codegen for TensorList (#73793)

wconstab committed 2 years ago

Verified 6537580c

Implement codegen repeat op (#73886)

wconstab committed 2 years ago

Verified 2ac8670c

Fix typo and broken link in LazyTensor documentation (#73861)

ansley committed 2 years ago

Verified b3ce386c

Fix rebase error

wconstab committed 2 years ago

5fcac745

delete unimplemented ops (nms, svd) (#73892)

wconstab committed 2 years ago

Verified 317700eb

[LT] Eliminate copy cost in the DDP eager fallback (#73729)

Jiewen Tan committed 2 years ago

Verified adf6c61f

Support _to_copy to enable AMP (#73613)

wconstab committed 2 years ago

Verified 2e79740f

Update lazy_bench.py (#73889)

wconstab committed 2 years ago

Verified 1d12b2ba

Implement TensorList and demonstrate it via Stack op (#73758)

wconstab committed 2 years ago

Verified 353aece7

use codegen for sub (#73615)

wconstab committed 2 years ago

Verified c28c7134

Disable special scalar handling for the TS backend

desertfire committed 2 years ago

7762e240

logging section in LTC (#73728)

Krovatkin committed 2 years ago

Verified c5ce0ef3

Update newly added _get_graph_hash and

desertfire committed 2 years ago

14a5479b

Use intrusive_ptr for LazyTensor (#73429)

wconstab committed 2 years ago

Verified 5571b476

LTC: reuse compiled graph (#72936)

shunting314 committed 2 years ago

Verified 9a288119

Merge branch 'wconstab/lazy_tensor_staging' into lazy_tensor_staging

wconstab committed 2 years ago

d9896b8b

Cherry-pick NVFuser's fix on return value number

desertfire committed 2 years ago

8457863d

Merge branch 'master' into wconstab/lazy_tensor_staging

wconstab committed 2 years ago

7f3db15e

Merge branch 'master' into wconstab/lazy_tensor_staging

wconstab committed 2 years ago

a0673f06

[qunat][fx][fix] Fix get_module_type for fusion (#72735)

jerryzh168 committed 2 years ago

16554bec

Revert D33992795: Add BUILD_LAZY_CUDA_LINALG option

janeyx99 committed 2 years ago

31271284

[FSDP][Reland] Implement local_state_dict and load_local_state_dict (#73300)

fegin committed 2 years ago

dc1bd971

[LT] Integrating with DDP with c10d comm ops falling back to eager (#72631)

Jiewen Tan committed 2 years ago

Verified bd383073

Fix more binary search overflow issues (#73239)

r-barnes committed 2 years ago

fe820b38

[quant][fx] Move MatchAllNode from match_utils.py to utils.py under quantization (#73344)

jerryzh168 committed 2 years ago

ee5b8f0c

[quant][graphmode] produce reference pattern for binary ops and then rewrite to quantized op (#72953)

jerryzh168 committed 2 years ago

9db0e0e7

Revert "Improve logging on ONNX symbolic helpers"

pytorchmergebot committed 2 years ago

b3c435be

Older