Commits

Branch selector

 main

User selector

bdhirsh

Datepicker

All time

Commit History

Commits on Dec 5, 2023

• remove aot_config.keep_inference_input_mutations from assert_functional_graph (pytorch#115195)

  Show description for 1102d37
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  1102d37

Commits on Dec 1, 2023

• AOTAutograd: support mutations on buffers that happen during the bw (pytorch#114953)

  Show description for c546ca9
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  c546ca9

Commits on Nov 29, 2023

• AOTAutograd: keep input mutations in the graph if they are under no_grad, even if they require_grad (pytorch#114646)

  Show description for 64ccdd4
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  64ccdd4

Commits on Nov 28, 2023

• [test] AOTAutograd: support mutations on buffers that happen during th bw (pytorch#112906)

  Show description for c8974d6
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  c8974d6

Commits on Nov 16, 2023

• graph break on intermediate leaves that require grad (pytorch#113277)

  Show description for cebad98
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  cebad98

Commits on Nov 15, 2023

• aot_autograd: keep input mutations on requires_grad=True tensor out of the graph for inference (pytorch#113584)

  Show description for cc11c0d
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  cc11c0d

• handle cross-dtype views during AOTAutograd view-replay (pytorch#113416)

  Show description for 032e5a4
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  032e5a4

• graph break on out= ops with noncontiguous out args (pytorch#113267)

  Show description for 720e866
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  720e866

Commits on Nov 13, 2023

• AOTAutograd: handle set_(), detect metadata mutations that cancel out (pytorch#111554)

  Show description for 3afb4e5
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  3afb4e5

Commits on Nov 10, 2023

• Fix selective activation checkpointing with subclasses that override sizes() (pytorch#113380)

  Show description for 7064fbf
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  7064fbf

Commits on Oct 26, 2023

• AOTAutograd: avoid intermediate_base logic when all aliased outputs came from a multi_output_view (pytorch#111411)

  Show description for 2aaa7e5
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  2aaa7e5

Commits on Oct 22, 2023

• python_arg_parser + dynamic shapes: fix segfault coercing symint to intlist (pytorch#111642)

  Show description for c9ca0dd
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  c9ca0dd

• dynamo: graph break on resize_ (pytorch#111553)

  Show description for 62942b0
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  62942b0

Commits on Oct 17, 2023

• Reland "AOTAutograd: Go down inference path if no outputs require grad (pytorch#111011)" (pytorch#111347)

  Show description for 7fb09b8
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  7fb09b8

Commits on Oct 16, 2023

• fix wrong meta for index_select.out (pytorch#111364)

  Show description for 0d368f5
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  0d368f5

• FunctionalTensor: avoid spurious not_implemented logging during proxy tracing (pytorch#111040)

  Show description for 4cf23c6
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  4cf23c6

Commits on Oct 11, 2023

• torch.compile DTensor E2E (pytorch#105236)

  Show description for 4d29b40
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  4d29b40

• AOTAutograd: Go down inference path if no outputs require grad (pytorch#111011)

  Show description for ded5ee7
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  ded5ee7

Commits on Oct 10, 2023

• AOTDispatch subclass (pytorch#104483)

  Show description for ba86dfc
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  ba86dfc

Commits on Oct 3, 2023

• Reland attempt 2 of "Update AOTAutograd to use FunctionalTensorMode instead of C++ functionalization (pytorch#106406)" (pytorch#109906)" (pytorch#110079)

  Show description for b457e3f
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  b457e3f

Commits on Sep 25, 2023

• Reland "Update AOTAutograd to use FunctionalTensorMode instead of C++ functionalization (pytorch#106406)" (pytorch#109906)

  Show description for d0fe8fa
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  d0fe8fa

Commits on Sep 22, 2023

• Update AOTAutograd to use FunctionalTensorMode instead of C++ functionalization (pytorch#106406)

  Show description for b5d6e83
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  b5d6e83

• Make FunctionalTensor subclass to be more like functorch (interaction with ZeroTensor + Conjugate key) (pytorch#109023)

  Show description for 63526a6
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  63526a6

• fix infinite loop with primtorch and .to(meta) (pytorch#109632)

  Show description for 7a21e96
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  7a21e96

• _return_and_correct_aliasing: fix for schemas with mutable tensor in kwargs (pytorch#109662)

  Show description for 46b0b7b
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  46b0b7b

• fix subclass custom sizes dynamic shapes caching (pytorch#108654)

  Show description for dae9aa8
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  dae9aa8

Commits on Sep 20, 2023

• python functionalization: support higher order ops (pytorch#108656)

  Show description for 238fb66
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  238fb66

• custom ops: don't error if autograd input is a tensor subclass (pytorch#109248)

  Show description for d9342cd
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  d9342cd

• functorch: fallthrough on calls to custom size/stride/storage_offset calls (pytorch#109024)

  Show description for c9b60a6
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  c9b60a6

Commits on Sep 19, 2023

• reland "python functionalization: add helpers, functionalize_sync and mirror_autograd_meta (pytorch#107917)" (pytorch#109518)

  Show description for 25e81f1
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  25e81f1

Commits on Sep 15, 2023

• return_and_correct_aliasing: massage some schemas to work with torchgen (pytorch#108897)

  Show description for 71b4b32
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  71b4b32

• python functionalization: add helpers, functionalize_sync and mirror_autograd_meta (pytorch#107917)

  Show description for 0ad5959
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  0ad5959

• Add TorchDispatch version of functionalization (pytorch#106404)

  Show description for f22b303
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  f22b303

Commits on Sep 5, 2023

• fix issue with lift_fresh_copy when using export + compile (pytorch#108243)

  Show description for fac7a1f
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  fac7a1f

• error when using _dynamo.optimize_ddp=True and _inductor.keep_output_stride=False together (pytorch#108235)

  Show description for da914ae
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  da914ae

Pagination

PreviousNext