strisunshinewentingwang/pytorch

Pull Requests Commits

[dtensor][debug] visualize_sharding skip if the current rank is not in mesh (#121382)

XilunWu committed 192 days ago

3a5ab17b

[dtensor][debug] visualize_sharding only compute offset on the first rank in mesh (#121385)

XilunWu committed 192 days ago

b383123e

Fix `get_rank` under a non-default group. (#120481)

KungYork committed 192 days ago

9c50ecc8

[dynamo] Fix support for nn.Parameter constructor (part 1) (#120163)

jansel committed 192 days ago

7cc476ea

[dynamo] Support _unsafe_set_version_counter (#121086)

jansel committed 192 days ago

32488b06

[Dynamo] Fix function overrides (#120885)

StellarrZ committed 193 days ago

7a4e4511

split predispatch pass into multiple passes (#121592)

kflu committed 193 days ago

f11f2b0d

relax assertion on fake shape (#121599)

avikchaudhuri committed 193 days ago

13e8181b

[Export] Fix bug removing node from wrong graph (#121574)

oulgen committed 193 days ago

660ec3d3

[IntraNodeComm] fix a hybridCubeMeshAllReduceKernel breakage caused by a recent refactor (#121575)

yifuwang committed 194 days ago

41286f15

[DeviceMesh] Add support for nD slicing (#119752)

wz337 committed 194 days ago

60cd2a43

[inductor] Log triton kernel source and metadata on failure (#120494)

bertmaher committed 194 days ago

e90cddb0

[inductor] Changes to support newer triton pin (#121267)

peterbell10 committed 194 days ago

168a04e7

[inductor] Refactor common triton imports into one function (#121438)

peterbell10 committed 194 days ago

459c5bca

Remove opmath cast for im2col decomp (#121363)

BowenBao committed 194 days ago

8c96b436

[dynamo] support rewriting dist.all_reduce with explicitly specified reduce op (#120181)

yifuwang committed 194 days ago

71d02026

Revert "Add CUTLASS kernel as choice for _int_mm() Inductor autotuning (#119685)"

pytorchmergebot committed 194 days ago

cf974237

[profiler] Fix recorded profiler step number (#121127)

fwenguang committed 194 days ago

761783a4

[dtensor] add async_op option to redistribute and some refactor (#121477)

wanchaol committed 194 days ago

242e03ba

[quant] Add error check for input_edge annotation (#121536)

jerryzh168 committed 194 days ago

a6a67da3

[export] Add effect token to export (#121424)

angelayi committed 194 days ago

e8836759

[C10d][NCCL] Refactor complex all_reduce and broadcast (#121045)

Aidyn-A committed 195 days ago

eb391994

Add CUTLASS kernel as choice for _int_mm() Inductor autotuning (#119685)

alexsamardzic committed 195 days ago

752d164b

[export] improve binary op fast path broadcast check (#121546)

ColinPeppler committed 195 days ago

13a25c64

[DCP] Makes fsspec public (#121508)

LucasLLC committed 195 days ago

d482614f

Change default torch_function behavior to be disabled when torch_dispatch is defined (take 2) (#120632)

albanD committed 195 days ago

6791b0c0

[CUDA graphs] Pool argument for make_graphed_callables (#121475)

Aidyn-A committed 195 days ago

ca967840

[C10d][UCC] Retain CUDA context in progress_loop (#121446)

Aidyn-A committed 195 days ago

b2f19dd2

Changed cublas repdocubility URL (#121534)

Chillee committed 195 days ago

ed8eebd1

[DCP] Replaced `storage()` with `untyped_storage()` (#121538)

awgu committed 195 days ago

b0a0850a

Newer Older