Commits

Branch selector

 main

User selector

wz337

Datepicker

All time

Commit History

Commits on Jul 25, 2024

• [DeviceMesh] Remove _parent_mesh as an attribute from DeviceMesh and remove it from DeviceMesh's hash (pytorch#131636)

  Show description for 4de85e3
  

  wz337

  authored and

  pytorchmergebot

  committed
  4de85e3

  Copy full SHA for 4de85e3

Commits on Jul 18, 2024

• [DTensor] Fix shard_dim_alltoall fake tensor return (pytorch#129945)

  Show description for 85451b2
  

  wz337

  authored and

  pytorchmergebot

  committed
  85451b2

  Copy full SHA for 85451b2

Commits on Jul 15, 2024

• [DeviceMesh][Reland] Only include the real thread_id in DeviceMesh hash under threaded backend (pytorch#130495) (pytorch#130685)

  Show description for ee6f0ab
  

  wz337

  authored and

  pytorchmergebot

  committed
  ee6f0ab

  Copy full SHA for ee6f0ab

Commits on Jul 11, 2024

• Change deprecated warning on dispatch_on_subclass to warn once (pytorch#130047)

  Show description for f21a218
  

  wz337

  authored and

  pytorchmergebot

  committed
  f21a218

  Copy full SHA for f21a218

• [DeviceMesh] Only include the real thread_id in DeviceMesh hash under threaded backend (pytorch#130495)

  Show description for 3896ba3
  

  wz337

  authored and

  pytorchmergebot

  committed
  3896ba3

  Copy full SHA for 3896ba3

Commits on Jul 5, 2024

• Change depreacate warning on dispatch_on_subclass to warn once (pytorch#130047)

  Show description for 8ff243b
  

  wz337

  authored and

  pytorchmergebot

  committed
  8ff243b

  Copy full SHA for 8ff243b

Commits on Jul 2, 2024

• [FSDP1][Easy] Remove Spammy Log Lin in _runtime_utils.py (pytorch#129967)

  Show description for 1ddb100
  

  wz337

  authored and

  pytorchmergebot

  committed
  1ddb100

  Copy full SHA for 1ddb100

• [FSDP2+TP] Disable 2D state_dict (pytorch#129519)

  Show description for 83e6ec2
  

  wz337

  authored and

  pytorchmergebot

  committed
  83e6ec2

  Copy full SHA for 83e6ec2

Commits on Jun 26, 2024

• [DCP] Fix Optimizer Learning Rate not being loaded correctly (pytorch#129398)

  Show description for 8b8e2fc
  

  wz337

  authored and

  pytorchmergebot

  committed
  8b8e2fc

  Copy full SHA for 8b8e2fc

Commits on Jun 25, 2024

• [DTensor][Test] Update implicit replication unit tests for tensor arg being the first in args list (pytorch#127803)

  Show description for d1f9e82
  

  wz337

  authored and

  pytorchmergebot

  committed
  d1f9e82

  Copy full SHA for d1f9e82

• [Test][DTensor] Temporarily skip gloo test for test_depthwise_convolution (pytorch#129391)

  Show description for 8b6391e
  

  wz337

  authored and

  pytorchmergebot

  committed
  8b6391e

  Copy full SHA for 8b6391e

Commits on Jun 21, 2024

• [FSDP][Test] Test save model save with FSDP1 and load into FSDP2 applied model (pytorch#129028)

  Show description for 31c9e3d
  

  wz337

  authored and

  pytorchmergebot

  committed
  31c9e3d

  Copy full SHA for 31c9e3d

Commits on Jun 8, 2024

• [DeviceMesh] Update get_group and add get_all_groups (pytorch#128097)

  Show description for 1d84c7e
  

  wz337

  authored and

  pytorchmergebot

  committed
  1d84c7e

  Copy full SHA for 1d84c7e

Commits on May 31, 2024

• [DeviceMesh] Adding nD slicing support back (pytorch#127465)

  Show description for 1699eda
  

  wz337

  authored and

  pytorchmergebot

  committed
  1699eda

  Copy full SHA for 1699eda

Commits on May 30, 2024

• [DeviceMesh] Adding nD slicing support back (pytorch#127465)

  Show description for e72232f
  

  wz337

  authored and

  pytorchmergebot

  committed
  e72232f

  Copy full SHA for e72232f

Commits on May 23, 2024

• [FSDP1] Update docstring to include device_mesh arg (pytorch#126589)

  Show description for 0d2ac97
  

  wz337

  authored and

  pytorchmergebot

  committed
  0d2ac97

  Copy full SHA for 0d2ac97

• [DCP][AC] Add test for apply AC with FSDP1 (pytorch#126935)

  Show description for 264155a
  

  wz337

  authored and

  pytorchmergebot

  committed
  264155a

  Copy full SHA for 264155a

Commits on May 21, 2024

• [DTensor][Optim] Add support for fused_adam and fused_adamw when lr is a tensor (pytorch#126750)

  Show description for fed536d
  

  wz337

  authored and

  pytorchmergebot

  committed
  fed536d

  Copy full SHA for fed536d

Commits on May 17, 2024

• [DTensor] Turn on foreach implementation for clip_grad_norm_ for DTensor by default (pytorch#126423)

  Show description for 15ca562
  

  wz337

  authored and

  pytorchmergebot

  committed
  15ca562

  Copy full SHA for 15ca562

• [FSDP2][Test] Fix _test_clip_grad_norm (pytorch#126457)

  Show description for 8858219
  

  wz337

  authored and

  pytorchmergebot

  committed
  8858219

  Copy full SHA for 8858219

Commits on May 16, 2024

• [DeviceMesh] Fix hash and eq not match (pytorch#123572)

  Show description for 059b68f
  

  wz337

  authored and

  pytorchmergebot

  committed
  059b68f

  Copy full SHA for 059b68f

Commits on May 15, 2024

• [DTensor] Turn on foreach implementation of optimizer for DTensor by default (pytorch#123394)

  Show description for c1dc8bb
  

  wz337

  authored and

  pytorchmergebot

  committed
  c1dc8bb

  Copy full SHA for c1dc8bb

Commits on May 8, 2024

• [DTensor] allow numel 1 tensor operand to be implicitly replicate DTensor (pytorch#125073)

  Show description for 603d1e6
  

  wz337

  authored and

  pytorchmergebot

  committed
  603d1e6

  Copy full SHA for 603d1e6

Commits on Apr 26, 2024

• [DeviceMesh] Removing mapping child_to_parent_mapping from _MeshEnv (pytorch#124890)

  Show description for 4ad291d
  

  wz337

  authored and

  pytorchmergebot

  committed
  4ad291d

  Copy full SHA for 4ad291d

• Making _MeshEnv subclassing thread local (pytorch#124555)

  Show description for 43f4e71
  

  wz337

  authored and

  pytorchmergebot

  committed
  43f4e71

  Copy full SHA for 43f4e71

Commits on Apr 24, 2024

• [DTensor][Easy] Update OpSchema __repr__ to show args_schema in format print (pytorch#124812)

  Show description for 7809b34
  

  wz337

  authored and

  pytorchmergebot

  committed
  7809b34

  Copy full SHA for 7809b34

• [FSDP1][2D] Fix FSDP1 2D state_dict to use run_check=False (pytorch#123802)

  Show description for 2716e77
  

  wz337

  authored and

  pytorchmergebot

  committed
  2716e77

  Copy full SHA for 2716e77

Commits on Apr 18, 2024

• [DeviceMesh][Test] Add 3d unit test for get_local_rank() (pytorch#124142)

  Show description for 2f45be4
  

  wz337

  authored and

  pytorchmergebot

  committed
  2f45be4

  Copy full SHA for 2f45be4

• [DTensor][Test] Add unit tests to keep track of DTensor sharding for 2D (pytorch#123687)

  Show description for 9fed2e8
  

  wz337

  authored and

  pytorchmergebot

  committed
  9fed2e8

  Copy full SHA for 9fed2e8

Commits on Apr 17, 2024

• [DeviceMesh][Test] Add 3d unit test for get_local_rank() (pytorch#124142)

  Show description for a403757
  

  wz337

  authored and

  pytorchmergebot

  committed
  a403757

  Copy full SHA for a403757

• [Test][2D] Turn on 2D state_dict tests for uneven sharding (pytorch#124255)

  Show description for cdc855a
  

  wz337

  authored and

  pytorchmergebot

  committed
  cdc855a

  Copy full SHA for cdc855a

Commits on Apr 10, 2024

• [DeviceMesh] Make dtype of mesh tensor from init_device_mesh() consistent with directly calling DeviceMesh() (pytorch#123677)

  Show description for 37fd547
  

  wz337

  authored and

  pytorchmergebot

  committed
  37fd547

  Copy full SHA for 37fd547

Commits on Mar 30, 2024

• [DeviceMesh] Cache and reuse sliced result (pytorch#122975)

  Show description for 2b1ba0c
  

  wz337

  authored and

  pytorchmergebot

  committed
  2b1ba0c

  Copy full SHA for 2b1ba0c

• [DCP][FSDP2][Test] Add_adamW to test_train_parity_2d_transformer_checkpoint_resume (pytorch#122002)

  Show description for 07f0ff6
  

  wz337

  authored and

  pytorchmergebot

  committed
  07f0ff6

  Copy full SHA for 07f0ff6

Commits on Mar 26, 2024

• [Easy][DCP] Fix small typo in assert (pytorch#122633)

  Show description for 8cfbdc0
  

  wz337

  authored and

  pytorchmergebot

  committed
  8cfbdc0

  Copy full SHA for 8cfbdc0

Pagination

Previous

Next