Code

PyTorch Softmax Ops#846

Code

Merged

ricardoV94 merged 51 commits intopymc-devs:mainpymc-devs/pytensor:mainfrom HAKSOAT:pytensor-pytorch-softmaxHAKSOAT/pytensor:pytensor-pytorch-softmaxCopy head branch name to clipboardJun 28, 2024

+100-1

Commits

Commits on May 13, 2024

• Add pytorch support for some basic Ops

  HarshvirSandhu

  committed
  27e2526

  Copy full SHA for 27e2526

• update variable names, docstrings

  HarshvirSandhu

  committed
  629d00b

  Copy full SHA for 629d00b

Commits on May 17, 2024

• Avoid numpy conversion of torch Tensors

  HarshvirSandhu

  committed
  3eceb56

  Copy full SHA for 3eceb56

• Fix typify and CheckAndRaise

  HarshvirSandhu

  committed
  3cde964

  Copy full SHA for 3cde964

• Fix Elemwise Ops

  HarshvirSandhu

  committed
  c003aa5

  Copy full SHA for c003aa5

• Fix Scalar Ops

  HarshvirSandhu

  committed
  8dc406e

  Copy full SHA for 8dc406e

• Fix ruff-format

  HarshvirSandhu

  committed
  a8f6ddb

  Copy full SHA for a8f6ddb

Commits on May 23, 2024

• Initial setup for pytorch tests

  HarshvirSandhu

  committed
  9d535f5

  Copy full SHA for 9d535f5

• Fix mode parameters for pytorch

  HarshvirSandhu

  committed
  c5600da

  Copy full SHA for c5600da

• Prevent conversion of scalars to numpy

  HarshvirSandhu

  committed
  54b6248

  Copy full SHA for 54b6248

• Update TensorConstantSignature and map dtypes to Tensor types

  HarshvirSandhu

  committed
  19454b3

  Copy full SHA for 19454b3

• Add tests for basic ops

  HarshvirSandhu

  committed
  92d7114

  Copy full SHA for 92d7114

Commits on May 29, 2024

• Remove torch from user facing API

  HarshvirSandhu

  committed
  5aae0e5

  Copy full SHA for 5aae0e5

• Add function to convert numpy arrays to pytorch tensors

  HarshvirSandhu

  committed
  8c174dd

  Copy full SHA for 8c174dd

• Avoid copy when converting to tensor

  HarshvirSandhu

  committed
  0977c3a

  Copy full SHA for 0977c3a

• Fix tests

  HarshvirSandhu

  committed
  1c23825

  Copy full SHA for 1c23825

Commits on May 31, 2024

• Remove dispatches that are not tested

  HarshvirSandhu

  committed
  c9195a8

  Copy full SHA for c9195a8

• set path for pytorch tests

  HarshvirSandhu

  committed
  b07805c

  Copy full SHA for b07805c

Commits on Jun 4, 2024

• Remove tensorflow probability from yml

  HarshvirSandhu

  committed
  9e8d3fc

  Copy full SHA for 9e8d3fc

• Add checks for runtime broadcasting

  HarshvirSandhu

  committed
  a2d3afa

  Copy full SHA for a2d3afa

• Remove IfElse

  HarshvirSandhu

  committed
  a577a80

  Copy full SHA for a577a80

Commits on Jun 12, 2024

• Remove dev notebook

  HarshvirSandhu

  committed
  499a174

  Copy full SHA for 499a174

• Fix check and raise

  HarshvirSandhu

  committed
  2826613

  Copy full SHA for 2826613

• Fix compare_pytorch_and_py

  HarshvirSandhu

  committed
  62ffcec

  Copy full SHA for 62ffcec

• Fix DimShuffle

  HarshvirSandhu

  committed
  acdbba1

  Copy full SHA for acdbba1

• Add tests for Elemwise operations

  HarshvirSandhu

  committed
  2519c65

  Copy full SHA for 2519c65

Commits on Jun 14, 2024

• Fix test for CheckAndRaise

  HarshvirSandhu

  committed
  eb6d5c2

  Copy full SHA for eb6d5c2

• Remove duplicate function

  HarshvirSandhu

  committed
  9f02a4f

  Copy full SHA for 9f02a4f

Commits on Jun 15, 2024

• Remove device from pytorch_typify

  HarshvirSandhu

  committed
  caf2965

  Copy full SHA for caf2965

• Merge branch 'main' of https://github.com/HarshvirSandhu/pytensor into pytensor-pytorch

  HarshvirSandhu

  committed
  bf87eb9

  Copy full SHA for bf87eb9

• Solve merge conflict

  HarshvirSandhu

  committed
  2c27683

  Copy full SHA for 2c27683

• Use micromamba for pytorch install

  HarshvirSandhu

  committed
  c603c6b

  Copy full SHA for c603c6b

Commits on Jun 16, 2024

• Fix pytorch linker

  HarshvirSandhu

  committed
  3f17107

  Copy full SHA for 3f17107

• Fix typify and deepcopy

  HarshvirSandhu

  committed
  e850d8d

  Copy full SHA for e850d8d

• Parametrize device in all tests

  HarshvirSandhu

  committed
  e682fc4

  Copy full SHA for e682fc4

• Install torch with cuda

  HarshvirSandhu

  committed
  bf4cf92

  Copy full SHA for bf4cf92

• Fix test_pytorch_FunctionGraph_once

  HarshvirSandhu

  committed
  899e7f9

  Copy full SHA for 899e7f9

• Remove device argument from test

  HarshvirSandhu

  committed
  04d2935

  Copy full SHA for 04d2935

Commits on Jun 17, 2024

• remove device from elemwise tests and add assertions

  HarshvirSandhu

  committed
  8ec7661

  Copy full SHA for 8ec7661

• skip tests if cuda is not available

  HarshvirSandhu

  committed
  bb7df41

  Copy full SHA for bb7df41

Commits on Jun 18, 2024

• Fix tests

  HarshvirSandhu

  committed
  0441cf2

  Copy full SHA for 0441cf2

Commits on Jun 20, 2024

• Merge branch 'main' of https://github.com/pymc-devs/pytensor into pytensor-pytorch-softmax

  HAKSOAT

  committed
  85f2742

  Copy full SHA for 85f2742

Commits on Jun 23, 2024

• Implemented softmax ops for PyTorch

  HAKSOAT

  committed
  4ca5aca

  Copy full SHA for 4ca5aca

• Merge remote-tracking branch 'upstream/main' into pytensor-pytorch-softmax

  HAKSOAT

  committed
  b9aca57

  Copy full SHA for b9aca57

Commits on Jun 24, 2024

• Switched to run softmax on all items if axis is None

  HAKSOAT

  committed
  287d9c2

  Copy full SHA for 287d9c2

• Implemented log softmax

  HAKSOAT

  committed
  f42e2a0

  Copy full SHA for f42e2a0

Commits on Jun 25, 2024

• Implemented softmaxgrad

  HAKSOAT

  committed
  35b17e0

  Copy full SHA for 35b17e0

Commits on Jun 27, 2024

• Added checks and error raises for nonfloat inputs

  HAKSOAT

  committed
  5efc3c8

  Copy full SHA for 5efc3c8

Commits on Jun 28, 2024

• Added checks and error raises for nonfloat inputs

  HAKSOAT

  committed
  16e415a

  Copy full SHA for 16e415a

• Merge branch 'pytensor-pytorch-softmax' of https://github.com/HAKSOAT/pytensor into pytensor-pytorch-softmax

  HAKSOAT

  committed
  ffbc594

  Copy full SHA for ffbc594

• Merge branch 'main' of https://github.com/pymc-devs/pytensor into pytensor-pytorch-softmax

  HAKSOAT

  committed
  b4cdce0

  Copy full SHA for b4cdce0