Commits

Branch selector

 main

User selector

davidberard98

All time

Commit History

Commits on May 31, 2024

• [NestedTensor] Use maybe_mark_dynamic instead of mark_dynamic (pytorch#127453)

  Show description for f33beb7
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  f33beb7

  Copy full SHA for f33beb7

Commits on May 17, 2024

• Use object identity for deepcopy memo (pytorch#126126)

  Show description for cb3b8cd
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  cb3b8cd

  Copy full SHA for cb3b8cd

Commits on May 14, 2024

• [profiler] Report strides in json trace (pytorch#125851)

  Show description for a00a99e
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  a00a99e

  Copy full SHA for a00a99e

Commits on May 13, 2024

• [NT] Make NestedTensor register as having symbolic sizes/strides (pytorch#124687)

  Show description for 82edc8b
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  82edc8b

  Copy full SHA for 82edc8b

Commits on May 10, 2024

• Add "accurate" FlopCounter implementations for NestedTensor SDPA kernels (pytorch#125776)

  Show description for 9e85d3d
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  9e85d3d

  Copy full SHA for 9e85d3d

Commits on May 9, 2024

• Fix common_methods_invocations example inputs to _efficient_attention_forward (pytorch#125788)

  Show description for 1ecea51
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  1ecea51

  Copy full SHA for 1ecea51

Commits on May 6, 2024

• Nested Tensor + AOTI test (pytorch#125513)

  Show description for 05bd7fe
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  05bd7fe

  Copy full SHA for 05bd7fe

Commits on May 1, 2024

• [inductor] switch assume_aligned_inputs to False (pytorch#124336)

  Show description for 9022f13
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  9022f13

  Copy full SHA for 9022f13

• [TorchScript] attach target function to OSError when source can't be found (pytorch#125248)

  Show description for e9ce239
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  e9ce239

  Copy full SHA for e9ce239

Commits on Apr 26, 2024

• Implement deepcopy / clone for SymNode, NestedIntSymNode (pytorch#121361)

  Show description for b3cf36c
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  b3cf36c

  Copy full SHA for b3cf36c

Commits on Apr 25, 2024

• [inductor] Specialize on unguarded alignment of example inputs (pytorch#123319)

  Show description for 4259e5d
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  4259e5d

  Copy full SHA for 4259e5d

Commits on Apr 12, 2024

• [Dispatcher] Collect autograd sequence numbers on PythonTLSSnapshot dispatch keys (pytorch#123304)

  Show description for 868e5ce
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  868e5ce

  Copy full SHA for 868e5ce

Commits on Apr 11, 2024

• [Tensor] Detect more cases of symbolic sizes/strides (pytorch#123696)

  Show description for 6b24ec4
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  6b24ec4

  Copy full SHA for 6b24ec4

Commits on Mar 28, 2024

• [docs] Update PT2+Profiler docs (pytorch#122272)

  Show description for 59f6393
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  59f6393

  Copy full SHA for 59f6393

Commits on Mar 22, 2024

• [inductor] config to control whether we assume inputs are aligned (pytorch#122158)

  Show description for 8013c44
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  8013c44

  Copy full SHA for 8013c44

Commits on Mar 14, 2024

• [jit] ClassType hashing: hash on compilation_unit as well (pytorch#121928)

  Show description for cceabe8
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  cceabe8

  Copy full SHA for cceabe8

• [jit] AliasDB type hash - don't always return 0 (pytorch#121874)

  Show description for 2d9cee2
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  2d9cee2

  Copy full SHA for 2d9cee2

• [jit] support getattr/hasattr on NamedTuple (pytorch#121863)

  Show description for 3df1b3b
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  3df1b3b

  Copy full SHA for 3df1b3b

Commits on Mar 8, 2024

• Revert "[fake_impls] Fix seed/offset device for attention kernels (pytorch#120839)" (pytorch#121447)

  Show description for 0f8c9ac
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  0f8c9ac

  Copy full SHA for 0f8c9ac

Commits on Mar 4, 2024

• [fake_impls] Fix seed/offset device for attention kernels (pytorch#120839)

  Show description for df3c8b8
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  df3c8b8

  Copy full SHA for df3c8b8

• [jit][nested strided tensor] support nested tensor in check_trace (pytorch#121039)

  Show description for cead036
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  cead036

  Copy full SHA for cead036

Commits on Mar 2, 2024

• Fix more xfails for scaled_dot_product_attention (pytorch#121032)

  Show description for 85157af
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  85157af

  Copy full SHA for 85157af

Commits on Mar 1, 2024

• Update xfails for scaled_dot_product_attention (pytorch#120928)

  Show description for 6ed2639
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  6ed2639

  Copy full SHA for 6ed2639

Commits on Feb 29, 2024

• [fake_impls] fix max_seqlen return values in efficient_attention_forward (pytorch#120842)

  Show description for df1e855
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  df1e855

  Copy full SHA for df1e855

Commits on Feb 28, 2024

• Move attention kernels from meta_registrations to fake_impls (pytorch#120682)

  Show description for d6c2029
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  d6c2029

  Copy full SHA for d6c2029

Commits on Feb 27, 2024

• [meta registration] efficient_attention_forward fix for NT inputs (pytorch#120594)

  Show description for fdae936
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  fdae936

  Copy full SHA for fdae936

Commits on Feb 20, 2024

• [aot_autograd] handle subclass input mutations correctly in collect_metadata_analysis.py (pytorch#120136)

  Show description for e54c4e8
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  e54c4e8

  Copy full SHA for e54c4e8

• [aot_autograd] support inplace mutations for subclasses (pytorch#120141)

  Show description for b364041
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  b364041

  Copy full SHA for b364041

Commits on Feb 8, 2024

• [NestedTensor] Support binary pointwise ops with >2 inputs (if inputs are non-tensors) (pytorch#119419)

  Show description for 278a0e1
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  278a0e1

  Copy full SHA for 278a0e1

Commits on Feb 1, 2024

• [dynamo] use par_style = "xar" in minifier targets file (pytorch#118603)

  Show description for 499f31d
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  499f31d

  Copy full SHA for 499f31d

Commits on Jan 31, 2024

• [meta registration] fix _efficient_attention_forward for jagged inputs (pytorch#118657)

  Show description for 1b03423
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  1b03423

  Copy full SHA for 1b03423

Commits on Jan 30, 2024

• [Nested Tensor] Support ragged_idx != 1 on aten::is_same_size, aten::_to_copy (pytorch#118442)

  Show description for 460950d
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  460950d

  Copy full SHA for 460950d

Commits on Jan 27, 2024

• [JIT] python IR bindings: consolidate tests, add short docs in OVERVIEW.md (pytorch#118319)

  Show description for 40c0879
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  40c0879

  Copy full SHA for 40c0879

Commits on Jan 26, 2024

• [Nested Tensor] view: basic support for ragged_idx != 1 and _unsafe_view (pytorch#118317)

  Show description for 2842d3c
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  2842d3c

  Copy full SHA for 2842d3c

Commits on Jan 25, 2024

• [NestedTensor] Support ragged_idx != 1 in pointwise ops (pytorch#118157)

  Show description for 52c5803
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  52c5803

  Copy full SHA for 52c5803

Pagination

Previous

Next