Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 15, 2025

• Fix score_mod.py dynamic max autotune for backward (#151270)

  Show description for ccfce9a
  

  fegin

  authored and

  pytorchmergebot

  committed
  ccfce9a

  Copy full SHA for ccfce9a

• [MPSInductor] Adjust memory format detection (#151288)

  Show description for afaadce
  

  malfet

  authored and

  pytorchmergebot

  committed
  afaadce

  Copy full SHA for afaadce

• [MPS] Fix logit output for half/bfloat (#151282)

  Show description for b8a2824
  

  malfet

  authored and

  pytorchmergebot

  committed
  b8a2824

  Copy full SHA for b8a2824

• [Dynamo] Fix the unimplemented_v2 of EventVariable.call_method in ctx_manager.py (#151208)

  Show description for a2f7764
  

  fffrog

  authored and

  pytorchmergebot

  committed
  a2f7764

  Copy full SHA for a2f7764

• make einsum unbacked friendly (#151032)

  Show description for 9e20a84
  

  ColinPeppler

  authored and

  pytorchmergebot

  committed
  9e20a84

  Copy full SHA for 9e20a84

• [cutlass backend] Add more logs for cutlass backend benchmark (#150639)

  Show description for 5a51de5
  

  henrylhtsang

  authored and

  pytorchmergebot

  committed
  5a51de5

  Copy full SHA for 5a51de5

• [c10d][fr] Enable FR analysis script for all fast-path coalesce op (#151243)

  Show description for 48b4bc1
  

  fduwjj

  authored and

  pytorchmergebot

  committed
  48b4bc1

  Copy full SHA for 48b4bc1

• [dynamo] Remove `traceable_tensor_subclasses`-related code (#151062)

  Show description for f66229d
  

  StrongerXi

  authored and

  pytorchmergebot

  committed
  f66229d

  Copy full SHA for f66229d

• [dynamo] handle tensor subclass with non-classmethod `__torch_function__` (#151061)

  Show description for 6a1499d
  

  StrongerXi

  authored and

  pytorchmergebot

  committed
  6a1499d

  Copy full SHA for 6a1499d

• [dynamo] Properly handle `super().some_classmethod(...)` (#151060)

  Show description for 73129b8
  

  StrongerXi

  authored and

  pytorchmergebot

  committed
  73129b8

  Copy full SHA for 73129b8

• clang-format CUDASymmetricMemory.cu (#151260)

  Show description for e178a3a
  

  Yifu Wang

  authored and

  pytorchmergebot

  committed
  e178a3a

  Copy full SHA for e178a3a

• Optimize typing in `lr_scheduler.py` (#151219)

  Show description for 25803d3
  

  zeshengzong

  authored and

  pytorchmergebot

  committed
  25803d3

  Copy full SHA for 25803d3

• test_store: fix timeout for test_queues (#151252)

  Show description for 4ede670
  

  d4l3k

  authored and

  pytorchmergebot

  committed
  4ede670

  Copy full SHA for 4ede670

• [PP] Add schedule visualizer (#150347)

  Show description for 263f08e
  

  H-Huang

  authored and

  pytorchmergebot

  committed
  263f08e

  Copy full SHA for 263f08e

Commits on Apr 14, 2025

• [MPSInductor] Fix silent correctness in bitcast (#151272)

  Show description for 070357b
  

  malfet

  authored and

  pytorchmergebot

  committed
  070357b

  Copy full SHA for 070357b

• [dynamo][invoke_subgraph] Use FxGraphModule comparison instead of hashing (#150911)

  Show description for 508b882
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  508b882

  Copy full SHA for 508b882

• [ROCm] Improve behavior of get_torch_rocm_version helper function on non-ROCm systems. (#151040)

  Show description for a24a9c4
  

  naromero77amd

  authored and

  pytorchmergebot

  committed
  a24a9c4

  Copy full SHA for a24a9c4

• Add inductor standalone_compile API (#150670)

  Show description for c9aef50
  

  oulgen

  authored and

  pytorchmergebot

  committed
  c9aef50

  Copy full SHA for c9aef50

• Revert "[map] always turn on dynamo for map (#150962)"

  Show description for 4a47dd9
  pytorchmergebot

  committed
  4a47dd9

  Copy full SHA for 4a47dd9

• Revert "[map] make proxy mode re-dispatch to fake key (#151034)"

  Show description for 6a77a0a
  pytorchmergebot

  committed
  6a77a0a

  Copy full SHA for 6a77a0a

• Mark auto_functionalized HOPs as cacheable (#151194)

  Show description for 070f389
  

  zou3519

  authored and

  pytorchmergebot

  committed
  070f389

  Copy full SHA for 070f389

• Improve sort with non-constant keys error message (#151193)

  Show description for dea50b0
  

  zou3519

  authored and

  pytorchmergebot

  committed
  dea50b0

  Copy full SHA for dea50b0

• [MPSInductor] Cast halfs to floats (#151246)

  Show description for 46ce8f7
  

  malfet

  authored and

  pytorchmergebot

  committed
  46ce8f7

  Copy full SHA for 46ce8f7

• Expand docs for `nn.functional`, and make the wording consistent (#148436)

  Show description for 0a6e1d6
  

  olipinski

  authored and

  pytorchmergebot

  committed
  0a6e1d6

  Copy full SHA for 0a6e1d6

• [c10d] Add `_allgather_base` , `reduce_scatter` , and `_reduce_scatter_base` into ProcessGroupMPI to enable FSDP with MPI backend (#150162)

  Show description for 23a3cef
  

  nariaki3551

  authored and

  pytorchmergebot

  committed
  23a3cef

  Copy full SHA for 23a3cef

• Fix assert_tensor_meta (#150808)

  Show description for 7deed19
  

  angelayi

  authored and

  pytorchmergebot

  committed
  7deed19

  Copy full SHA for 7deed19

• Generate meta kernel with operator profiles (#150807)

  Show description for 5352844
  

  angelayi

  authored and

  pytorchmergebot

  committed
  5352844

  Copy full SHA for 5352844

• [ONNX] Fix bfloat16 support in onnx_program callable (#151121)

  Show description for 901e375
  

  2 peopleauthored and

  pytorchmergebot

  committed
  901e375

  Copy full SHA for 901e375

• Add error check for out variant of tensordot function with requries_grad tensor (#150270)

  Show description for f76b7ef
  

  cz2h

  authored and

  pytorchmergebot

  committed
  f76b7ef

  Copy full SHA for f76b7ef

• Using hasattr for `_boxed_call` is asking for trouble (#151130)

  Show description for 1f5af12
  

  aorenste

  authored and

  pytorchmergebot

  committed
  1f5af12

  Copy full SHA for 1f5af12

• [dynamic shapes] add sym_and, sym_or (#150456)

  Show description for 6dddd65
  

  pianpwk

  authored and

  pytorchmergebot

  committed
  6dddd65

  Copy full SHA for 6dddd65

• [dynamo][error message] Hint for dict_items as inputs to the compiled region (#151169)

  Show description for 785495e
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  785495e

  Copy full SHA for 785495e

• [dynamo] Graph break fixes while tracing inspect module (#151168)

  Show description for 3c46808
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  3c46808

  Copy full SHA for 3c46808

• [scan] Autograd with partial gradient support (#146285)

  Show description for b0bdd76
  

  2 peopleauthored and

  pytorchmergebot

  committed
  b0bdd76

  Copy full SHA for b0bdd76

• Super tiny fix typo (#151212)

  Show description for 50abc1e
  

  fzyzcjy

  authored and

  pytorchmergebot

  committed
  50abc1e

  Copy full SHA for 50abc1e

Pagination

Previous

Next