strisunshinewentingwang/pytorch

Pull Requests Commits

[inductor] Stop using `x + tl.zeros(...)` in generated triton (#100163)

peterbell10 committed 1 year ago

5b989101

[inductor] Move reduction_type special cases out of make_reduction (#99660)

peterbell10 committed 1 year ago

270a3316

File level rerun changes (#100200)

clee2000 committed 1 year ago

6ab9453e

[CUDA] Switch to `at::empty_like` in `adaptive_avg_pool3d_backward_cuda` (#100202)

eqy committed 1 year ago

43dea763

Revert "Added round_with_scale_factor arg to ATen (#97868)"

pytorchmergebot committed 1 year ago

380ccfd4

Bump cuDNN frontend submodule to 0.9 (#99674)

eqy committed 1 year ago

5022143f

[CUDA] Do accumulation for Adaptive Average Pooling in `opmath_t` (#99378)

eqy committed 1 year ago

3f656ad7

Add aten.smooth_l1_loss_backward to core_aten_decompositions (#100267)

desertfire committed 1 year ago

b66d7007

Use `[[maybe_unused]]` in `VariableType_[0-4].cpp` (#100250)

crcrpar committed 1 year ago

9e1f46d5

[PG Wrapper] Enhance error msg (#100213)

rohan-varma committed 1 year ago

c4bed869

[SPMD] Introduce prerequisites to graph_optimization_pass (#99970)

fegin committed 1 year ago

e0a2b49f

[fix] masked_scatter_: non-contiguous self (#100232)

kshitij12345 committed 1 year ago

61dffa61

Add warning information for dtypetensor. (#99521)

NmomoN committed 1 year ago

9cd48b05

Pin functorch docs requirements (#100257)

huydhn committed 1 year ago

56e235ad

[c10d] Comment out ddp_hook_with_optimizer_parity tests (#100215)

kwen2501 committed 1 year ago

628a8df1

Fix triton auto update pin workflow (#100211)

clee2000 committed 1 year ago

efed5e1c

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

Newer Older