Commits

Branch selector

 main

User selector

janeyx99

Datepicker

All time

Commit History

Commits on Jul 10, 2023

• [foreach][AdamW] Minimize intermediates to save peak memory (pytorch#104898)

  Show description for 7e9c891
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  7e9c891

• [foreach][Adam] Minimize use of intermediates to decrease peak memory (pytorch#104780)

  Show description for 35f0e35
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  35f0e35

• Add meta registrations and distributed decomps: _foreach_div_.Scalar, sqrt_.default (pytorch#104779)

  Show description for e25f573
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  e25f573

• Add capturable/maximize tests to Adam(W) optim configs (pytorch#104669)

  Show description for 038cb40
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  038cb40

Commits on Jul 8, 2023

• [optim] use lerp whenever possible (pytorch#104796)

  Show description for fbe2a7e
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  fbe2a7e

Commits on Jun 13, 2023

• Fix optim state_dict casting to allow step to cast to CPU (pytorch#102619)

  Show description for fa893f3
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  fa893f3

Commits on Jun 9, 2023

• Disable dynamo'd test_optim entirely (pytorch#103323)

  Show description for 4a5d56b
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  4a5d56b

Commits on May 17, 2023

• [docs] Clarify when to use SparseAdam (pytorch#101465)

  Show description for 3135bec
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  3135bec

• Reorganize foreach ops more logically in native_functions.yaml (pytorch#101583)

  Show description for 7ca5e68
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  7ca5e68

Commits on May 16, 2023

• [docs] Warn that GradScaler can scale under 1 (pytorch#101569)

  Show description for cde597e
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  cde597e

Commits on May 12, 2023

• [BE] super small comment update to gradcheck.py (pytorch#101103)

  Show description for 4a7ee79
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  4a7ee79

• [optim][BE] split test file into logical parts: SWA, LR, optim (pytorch#101100)

  Show description for a53cda1
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  a53cda1

Commits on May 11, 2023

• [BE] Simplify tests, elaborate testnames in test_optim.py (pytorch#101004)

  Show description for cb94ea6
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  cb94ea6

Commits on May 9, 2023

• [optim] More cleanup and reorg of test_optim.py (pytorch#100917)

  Show description for d63e0b1
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  d63e0b1

• [BE][optim] Remove objects from being globals and comment to clarify (pytorch#100899)

  Show description for d0dab77
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  d0dab77

• [adam] Use the right params in weight_decay, rename for clarity, fixes pytorch#100707 (pytorch#100973)

  Show description for f558af2
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  f558af2

Commits on May 8, 2023

• [BE] Update notes linkage in common_device_type, fix very minor grammar (pytorch#100898)

  Show description for 9a811d1
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  9a811d1

Commits on Apr 25, 2023

• Prevent grad scale from overflowing (pytorch#98876)

  Show description for 8082677
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  8082677

Commits on Mar 20, 2023

• Allow fused optimizers to call _foreach_zero_ in zero_grad (pytorch#97159)

  Show description for aacbf09
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  aacbf09

Commits on Mar 17, 2023

• Make keep-going work for more than linux (pytorch#96974)

  Show description for 4a90aca
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  4a90aca

Commits on Mar 4, 2023

• [BE] Remind people there are rsts to update in docs/source (pytorch#95914)

  Show description for f4b3379
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  f4b3379

Commits on Mar 2, 2023

• [optim] Widen the cases for defaulting to foreach (pytorch#95820)

  Show description for 75cb99e
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  75cb99e

• [optim] include nn.Parameter as foreach supported (pytorch#95811)

  Show description for 2bcf863
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  2bcf863

Commits on Feb 28, 2023

• Rephrase zero_grad docs (pytorch#95643)

  Show description for e5b9d98
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  e5b9d98

Commits on Feb 24, 2023

• [optim] Add general documentation on our algorithm defaults (pytorch#95391)

  Show description for b215af2
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  b215af2

Commits on Feb 23, 2023

• Update docs that Parameters are immune to no_grad mode (pytorch#95232)

  Show description for 6dc81f7
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  6dc81f7

• Update docs that Parameters are immune to no_grad mode (pytorch#95232)

  Show description for 5783cee
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  5783cee

Commits on Feb 22, 2023

• [optim] Set defaults to foreach, NOT fused (pytorch#95241)

  Show description for 0976794
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  0976794

Commits on Feb 21, 2023

• [BE][CI] remove .jenkins entirely (pytorch#92625)

  Show description for 5d1fec8
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  5d1fec8

Commits on Jan 30, 2023

• [BE][CI] change references from .jenkins to .ci (pytorch#92624)

  Show description for 0ecb071
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  0ecb071

• [optim][adam] use fastest impl whenever possible, add util (pytorch#93184)

  Show description for 4fc19e1
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  4fc19e1

Commits on Jan 26, 2023

• [optim][sgd] default to foreach when CUDA + differentiable=False (pytorch#92730)

  Show description for e714e37
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  e714e37

• [foreach] guard default support on native tensors only (pytorch#92923)

  Show description for 8c9f745
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  8c9f745

• [nn] add set_to_none flag for C++ optim endpoint (pytorch#92989)

  Show description for 819bd5b
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  819bd5b

• [nn] zero_grad() set_to_none default True (pytorch#92731)

  Show description for b90496e
  

  janeyx99

  authored and

  pytorchmergebot

  committed
  b90496e

Pagination

Previous

Next