strisunshinewentingwang/pytorch

Pull Requests Commits

Improved CustomOp API with schema inference (#100127)

zou3519 committed 1 year ago

1b84be55

[CustomOp] Fix lifetime semantics (#100114)

zou3519 committed 1 year ago

7ebb60c9

[quant][pt2] Add test for prepare_qat Conv + BN numerics (#99846)

andrewor14 committed 1 year ago

d176e3ff

[Dynamo] Fix staticmethods for FSDP (#100117)

awgu committed 1 year ago

23de2e06

CustomOp simple abstract implementation registration (#99439)

zou3519 committed 1 year ago

e6f9bc50

Excise yaml dependency in torchgen.model (#100203)

zou3519 committed 1 year ago

4135295a

[inductor] Use decomposition for smooth_l1_loss_backward (#100242)

desertfire committed 1 year ago

55b66113

Enable test_linalg_solve_triangular_large (#96182)

iremyux committed 1 year ago

25040893

Revert "[CI] Start to collect inference perf with cpp_wrapper ON (#100187)"

pytorchmergebot committed 1 year ago

90c44b13

Add vmap support for `smooth_l1_loss_backward` (#99429)

yhl48 committed 1 year ago

07c02b9e

Persist torch.assert in aten graph (#100101)

tugsbayasgalan committed 1 year ago

d4bf76c2

[ROCm] Also look for 'Cijk' (rocblas kernel) to verify gemm in test_kineto (#92889)

jithunnair-amd committed 1 year ago

cef15ecc

Add experimental export() API (#100034)

gmagogsfm committed 1 year ago

751c54b5

[FSDP] Make set_state_type to SHARDED_STATE_DICT compatible with NO_SHARD sharding_strategy (#100208)

wz337 committed 1 year ago

a2336588

fix missing-prototypes warnings in torch_cpu (Part 2) (#100147)

cyyever committed 1 year ago

7220201a

Track exact origin_node on best effort basis (#100110)

ezyang committed 1 year ago

54c0edf6

[Tensor Parallel] Add a new Colwise Parallel style when Pairwise cannot directly used (#100137)

fduwjj committed 1 year ago

89b1e67d

Store constraints and example inputs in the graph module as metadata in export (#99961)

guangy10 committed 1 year ago

56a93ed5

enable some cuda warnings (#95568)

cyyever committed 1 year ago

c8877e60

[ONNX] Add xfail into subtests of op consistency and retire fixme (#100173)

titaiwangms committed 1 year ago

cba07ffe

Remove unecessary <execinfo.h> include (#99800)

doughdemon committed 1 year ago

6168bed6

[philox_rand] Add decomps (#100206)

anijain2305 committed 1 year ago

a8ad0dc3

[hotfix] Do not import torch.ao.quantization._pt2e from dynamo (#100194)

andrewor14 committed 1 year ago

9cda7b9e

Revert "[c10d] Comment out ddp_hook_with_optimizer_parity tests (#100215)"

pytorchmergebot committed 1 year ago

9609aeef

Improved message to suppress errors in _dynamo/exc.py (#97345)

vfdev-5 committed 1 year ago

0692bdd9

[Docs] Fix docstring format (#99396)

zhouzaida committed 1 year ago

b51f92eb

Add directly referenced header files for "ceil_div.h" (#99607)

caizhi-mt committed 1 year ago

64efd888

[CI] Start to collect inference perf with cpp_wrapper ON (#100187)

desertfire committed 1 year ago

3e87fc52

per-Tensor `grad_fn` for in-place foreach functions (#96405)

crcrpar committed 1 year ago

67401890

More rigorous mixed overloads on SymInt (#100008)

ezyang committed 1 year ago

9a3e411a

Newer Older