Repository navigation

Commits

Commits on Aug 17, 2024
- [compiled autograd] move non-hot path logs into default logger (#133541)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  983bea3
- [compiled autograd] use same graph node names as AOTDispatcher (#133148)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  0a6cc15
- [compiled autograd] log aot id for CompiledFunctionBackward (#133115)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  4b3ed8b
Commits on Aug 10, 2024
- [compiled autograd][cpp node] No recaptures from saved float scalars (#133048)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  b9922f7
- [compiled autograd][cpp node] No recompiles from saved int scalars (#132771)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  c860889
Commits on Jul 8, 2024
- [compiled autograd] c++ autograd function saved_data: lift tensors (#130057)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  010009e
Commits on Jul 3, 2024
- [dynamic] config to disable duck sizing (#129804)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  be2d79a
Commits on Jun 21, 2024
- [compiled autograd] flatten runtime inputs with fast path (#129116)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  f0443ad
- [compiled autograd] move inputs to cuda with non_blocking=True (#129181)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  d97dfe9
- [compiled autograd] treat input params as static (#128987)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  8f320fd
- [compiled autograd] use in_compiled_autograd_region instead of compiled_autograd_enabled_count (#128982)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  fafa186
- [aot autograd] collect static parameter metadata when graphs fallback to inference (#128905)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  68b3345
- [compiled autograd] update benchmarks to use cli flags for fullgraph/dynamic (#127960)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  1238127
Commits on Jun 14, 2024
- [dynamo][aot autograd] Silently disable default saved tensor hooks during tracing (#123196)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  4b96575
Commits on Jun 10, 2024
- [compiled autograd] support .backward(inputs=) (#128252)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  2176ef7
Commits on Jun 7, 2024
- [compiled autograd][cudagraphs] Inputs runtime wrapper to move cpu scalars to cuda (#125382)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  00c6ca4
Commits on May 31, 2024
- [compiled autograd] torch.compile API (#125880)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  ec098b8
Commits on May 30, 2024
- [compiled autograd] torch.compile API (#125880)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  e1c3221
Commits on May 22, 2024
- [dynamo][flaky tests] test_conv_empty_input_* (#126790)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  7530cfe
Commits on May 19, 2024
- [compiled autograd] Better cache miss logging (#126602)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  5fb11cd
- [compiled autograd] log in cpp using python logger (#126483)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  be67985
- Update hf_BirdBird periodic-dynamo-benchmarks results (#126414)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  5ea956a
Commits on May 16, 2024
- [compiled autograd] clear compiled_autograd_verbose once test is done (#126148)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  93524cf
- [inductor] Clear cache on ctx manager exit (#126146)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  cef7756
- [compiled autograd] Fix LoggingTensor flaky test (#126144)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  4cd4463
Commits on May 8, 2024
- [compiled autograd][dynamo] improve lifted autograd.Function.backward handling and fallback to pseudo-eager (#125661)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  7e0edaf
Commits on Apr 30, 2024
- Revert "[benchmark][cudagraph] Explicitly call aten.div with CUDA denominator for cudagraphs (#119729)" (#125246)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  c12c85e
Commits on Apr 27, 2024
- [compiled autograd] verbose logs for debugging cache misses (#124980)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  1a6fef1
- [compiled autograd] introduce verbose logs, add autograd node info to graph (#124954)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  43a7ab2
Commits on Apr 26, 2024
- [cudagraphs] add cudagraph_skips counter (#124804)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  1443056
- [cudagraphs] add more info to skip messages (#124700)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  8559399
- [benchmark][cudagraph] Explicitly call aten.div with CUDA denominator for cudagraphs (#119729)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  62b5738
Commits on Apr 25, 2024
- [cudagraphs] add cudagraph_skips counter (#124804)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  fdad16b
- [cudagraphs] add more info to skip messages (#124700)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  0ed38c9
- [benchmark][cudagraph] Explicitly call aten.div with CUDA denominator for cudagraphs (#119729)
  
  xmfan
  authored and
  pytorchmergebot
  committed
  c021c9b

Branch selector

User selector

Datepicker

Commit history

Commits on Aug 17, 2024

Commits on Aug 10, 2024

Commits on Jul 8, 2024

Commits on Jul 3, 2024

Commits on Jun 21, 2024

Commits on Jun 14, 2024

Commits on Jun 10, 2024

Commits on Jun 7, 2024

Commits on May 31, 2024

Commits on May 30, 2024

Commits on May 22, 2024

Commits on May 19, 2024

Commits on May 16, 2024

Commits on May 8, 2024

Commits on Apr 30, 2024

Commits on Apr 27, 2024

Commits on Apr 26, 2024

Commits on Apr 25, 2024