Commits

Breadcrumbs

History for

onremove-sentence-in-learnable-quantization-documentation

User selector

All users

Datepicker

All time

Commit History

Commits on May 8, 2024

• [dynamo] don't LOAD_FAST local context variables in modified bytecode (#125719)

  Show description for 255a3af
  

  williamwen42

  authored and

  pytorchmergebot

  committed
  255a3af

  Copy full SHA for 255a3af

Commits on May 5, 2024

• Also pull size/stride info from example_value (#125505)

  Show description for ecd6274
  

  ezyang

  authored and

  pytorchmergebot

  committed
  ecd6274

  Copy full SHA for ecd6274

Commits on May 4, 2024

• [dynamo] handle inactive nullcontexts across graph breaks (#125518)

  Show description for 1a20b4e
  

  williamwen42

  authored and

  pytorchmergebot

  committed
  1a20b4e

  Copy full SHA for 1a20b4e

Commits on May 3, 2024

• [dynamo] fix crash when context manager is passed to a function (#125321)

  Show description for f2ab96a
  

  williamwen42

  authored and

  pytorchmergebot

  committed
  f2ab96a

  Copy full SHA for f2ab96a

Commits on May 1, 2024

• [dynamo] support inactive context managers across graph breaks (#125203)

  Show description for 0506e95
  

  williamwen42

  authored and

  pytorchmergebot

  committed
  0506e95

  Copy full SHA for 0506e95

Commits on Apr 6, 2024

• [Compile FSDP2][2/n] Support streams created outside of compile region (#123487)

  Show description for 7b02910
  

  yf225

  authored and

  pytorchmergebot

  committed
  7b02910

  Copy full SHA for 7b02910

Commits on Feb 6, 2024

• [dynamo] support comparing stream with constant (#119199)

  Show description for 5410385
  

  ydwu4

  authored and

  pytorchmergebot

  committed
  5410385

  Copy full SHA for 5410385

Commits on Jan 26, 2024

• Deflake stream tests, part 2 (#118391)

  Show description for 728789d
  

  zou3519

  authored and

  pytorchmergebot

  committed
  728789d

  Copy full SHA for 728789d

Commits on Jan 24, 2024

• Deflake Dynamo stream tests (#118205)

  Show description for 6186520
  

  zou3519

  authored and

  pytorchmergebot

  committed
  6186520

  Copy full SHA for 6186520

Commits on Jan 23, 2024

• Replace invoking self.value if there is a user defined init, avoiding arbitrary code execution (#117818)

  Show description for fed45ae
  

  voznesenskym

  authored and

  pytorchmergebot

  committed
  fed45ae

  Copy full SHA for fed45ae

Commits on Jan 13, 2024

• Reconstruct streams via global registration, temporary impl to unblock FSDP (#117386)

  Show description for f008efa
  

  voznesenskym

  authored and

  pytorchmergebot

  committed
  f008efa

  Copy full SHA for f008efa

Commits on Jan 4, 2024

• Re-add initial Flash Attention support on ROCM (#115981)

  Show description for e3ca734
  

  xinyazhang

  authored and

  pytorchmergebot

  committed
  e3ca734

  Copy full SHA for e3ca734

Commits on Nov 30, 2023

• [Dynamo] Support torch.amp.autocast as decorator (#114845)

  Show description for 7f40640
  

  yanboliang

  authored and

  pytorchmergebot

  committed
  7f40640

  Copy full SHA for 7f40640

Commits on Nov 9, 2023

• [contextlib] Wrapping a function with `set_grad_enabled` will consume its global mutation (#113359)

  Show description for 5ccd225
  

  jon-chuang

  authored and

  pytorchmergebot

  committed
  5ccd225

  Copy full SHA for 5ccd225

• [dynamo] GradModeVariable should only be eagerly initialized when doing the equivalent of `set_grad_enabled` (#113293)

  Show description for 0093e23
  

  jon-chuang

  authored and

  pytorchmergebot

  committed
  0093e23

  Copy full SHA for 0093e23

Commits on Oct 22, 2023

• [dynamo] Fix context wrapping grad mode variable (#111534)

  Show description for 5af97fe
  

  jon-chuang

  authored and

  pytorchmergebot

  committed
  5af97fe

  Copy full SHA for 5af97fe

• [dynamo][stream]support device-agnostic stream in dynamo and capture stream/event method in fx graph (#108312)

  Show description for 8e60d64
  

  zejun-chen

  authored and

  pytorchmergebot

  committed
  8e60d64

  Copy full SHA for 8e60d64

Commits on Sep 13, 2023

• Flash Attention v2 (#105602)

  Show description for ad90ab3
  

  drisspg

  authored and

  pytorchmergebot

  committed
  ad90ab3

  Copy full SHA for ad90ab3

• Use user directed names for variables where possible (#109092)

  Show description for de0b18f
  

  voznesenskym

  authored and

  pytorchmergebot

  committed
  de0b18f

  Copy full SHA for de0b18f

Commits on Sep 8, 2023

• Revert "Flash Attention v2 (#105602)" (#108827)

  Show description for a9c663c
  

  huydhn

  authored and

  pytorchmergebot

  committed
  a9c663c

  Copy full SHA for a9c663c

• Revert "Revert "Flash Attention v2 (#105602)" (#108827)"

  Show description for e45b290
  pytorchmergebot

  committed
  e45b290

  Copy full SHA for e45b290

• Revert "Flash Attention v2 (#105602)" (#108827)

  Show description for 24e9bbe
  

  huydhn

  authored and

  pytorchmergebot

  committed
  24e9bbe

  Copy full SHA for 24e9bbe

Commits on Sep 1, 2023

• Flash Attention v2 (#105602)

  Show description for add45ae
  

  drisspg

  authored and

  pytorchmergebot

  committed
  add45ae

  Copy full SHA for add45ae

Commits on Aug 19, 2023

• [dynamo] fix disable_saved_tensors_hooks - graph break (#106875)

  Show description for 11602ac
  

  kshitij12345

  authored and

  pytorchmergebot

  committed
  11602ac

  Copy full SHA for 11602ac

Commits on Jul 27, 2023

• Change _dynamo.export to be export(f)(*args, **kwargs) (#106109)

  Show description for 7b9d250
  

  ezyang

  authored and

  pytorchmergebot

  committed
  7b9d250

  Copy full SHA for 7b9d250

Commits on Jul 26, 2023

• dynamo: support disable_saved_tensors_hooks (#104869)

  Show description for 920b446
  

  kshitij12345

  authored and

  pytorchmergebot

  committed
  920b446

  Copy full SHA for 920b446

Commits on Jun 26, 2023

• REDO of dropout support for mem eff #102038 (#103704)

  Show description for 4a008d2
  

  drisspg

  authored and

  pytorchmergebot

  committed
  4a008d2

  Copy full SHA for 4a008d2

Commits on Jun 15, 2023

• Switch dynamic_shapes to True by default (#103597)

  Show description for bc6ec97
  

  ezyang

  authored and

  pytorchmergebot

  committed
  bc6ec97

  Copy full SHA for bc6ec97

Commits on Jun 14, 2023

• Don't test dynamic_shapes in tensor_always_has_static_shape (#103517)

  Show description for 8b015c1
  

  ezyang

  authored and

  pytorchmergebot

  committed
  8b015c1

  Copy full SHA for 8b015c1

Commits on May 30, 2023

• [Dynamo] Save global autocast state to restore on graph break (#102415)

  Show description for 9ff1932
  

  yanboliang

  authored and

  pytorchmergebot

  committed
  9ff1932

  Copy full SHA for 9ff1932

Commits on May 24, 2023

• separate out dynamo .requires_grad and .is_grad_enabled guards (#100570)

  Show description for 98ab11a
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  98ab11a

  Copy full SHA for 98ab11a

• Disable (Broken) CUDAStreamVariable in dynamo (#100766)

  Show description for d55aad1
  

  voznesenskym

  authored and

  pytorchmergebot

  committed
  d55aad1

  Copy full SHA for d55aad1

Commits on May 19, 2023

• Revert "separate out dynamo .requires_grad and .is_grad_enabled guards (#100570)"

  Show description for 7f3fed1
  pytorchmergebot

  committed
  7f3fed1

  Copy full SHA for 7f3fed1

• separate out dynamo .requires_grad and .is_grad_enabled guards (#100570)

  Show description for 1fabee3
  

  bdhirsh

  authored and

  pytorchmergebot

  committed
  1fabee3

  Copy full SHA for 1fabee3

Commits on May 18, 2023

• [dynamo][moco] Save global torch state to restore on graph break (#101201)

  Show description for dafa009
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  dafa009

  Copy full SHA for dafa009

Pagination

Previous

Next