Commits

Branch selector

 main

User selector

zou3519

All time

Commit History

Commits on Feb 11, 2026

• [torch.compile] Stop doing unnecessary FakeTensorProp in PiecewiseCompileInterpreter (#34093)

  Show description for e30cedd
  zou3519

  authored
  e30cedd

  Copy full SHA for e30cedd

Commits on Feb 10, 2026

• [torch.compile] Disable recursive pre_grad_passes (#34092)

  Show description for 341eed3
  zou3519

  authored
  341eed3

  Copy full SHA for 341eed3

Commits on Feb 8, 2026

• [torch.compile] Add an option to force-enable the MOE cold start optimization (#33735)

  Show description for 4df841f
  zou3519

  authored
  4df841f

  Copy full SHA for 4df841f

Commits on Feb 7, 2026

• [torch.compile] Stop compiling identical artifacts (#34003)

  Show description for 81fe69c
  zou3519

  authored
  81fe69c

  Copy full SHA for 81fe69c

Commits on Feb 4, 2026

• Revert "[torch.compile] Significantly speed up cold start times" (#33820)

  Show description for 9f14c92
  zou3519

  authored
  9f14c92

  Copy full SHA for 9f14c92

Commits on Feb 3, 2026

• [torch.compile] Significantly speed up cold start times (#33641)

  Show description for b1bb18d
  zou3519

  authored
  b1bb18d

  Copy full SHA for b1bb18d

• [torch.compile] Document the workaround to standalone_compile failing (#33571)

  Show description for fd9c83d
  zou3519

  authored
  fd9c83d

  Copy full SHA for fd9c83d

• [torch.compile] Don't do the fast moe cold start optimization if there is speculative decoding (#33624)

  Show description for 5eac9a1
  

  zou3519

  and

  robertgshaw2-redhat

  authored
  5eac9a1

  Copy full SHA for 5eac9a1

Commits on Jan 27, 2026

• [torch.compile] Speed up MOE handling in forward_context (#33184)

  Show description for d9aa39a
  zou3519

  authored
  d9aa39a

  Copy full SHA for d9aa39a

• [torch.compile] Stop assuming 32 bit indexing (#33113)

  Show description for 3b8f0fe
  zou3519

  authored
  3b8f0fe

  Copy full SHA for 3b8f0fe

Commits on Jan 22, 2026

• [torch.compile] Improve Cold Start for MoEs (#32805)

  Show description for 654a71f
  zou3519

  authored
  654a71f

  Copy full SHA for 654a71f

Commits on Jan 15, 2026

• [BugFix] Python file source reading can fail on UnicodeDecodeError (#32416)

  Show description for bd292be
  zou3519

  authored
  bd292be

  Copy full SHA for bd292be

Commits on Jan 8, 2026

• [BugFix] Fix flakiness in test_eagle_dp for PyTorch 2.10 (#31915)

  Show description for a79079f
  zou3519

  authored
  a79079f

  Copy full SHA for a79079f

Commits on Dec 25, 2025

• [Bugfix] Fix eagle dp tests on A100 (#31241)

  Show description for 254f6b9
  zou3519

  authored
  254f6b9

  Copy full SHA for 254f6b9

Commits on Nov 5, 2025

• [Docs] Add guide to debugging vLLM-torch.compile integration (#28094)

  Show description for 65ac8d8
  zou3519

  authored
  65ac8d8

  Copy full SHA for 65ac8d8

Commits on Oct 24, 2025

• [compile] Turn standalone_compile back on (#27460)

  Show description for cd390b6
  zou3519

  authored
  cd390b6

  Copy full SHA for cd390b6

Commits on Oct 16, 2025

• [BugFix] Work around graph partition x torch.compile cache issue (#26956)

  Show description for 9b6504c
  zou3519

  authored
  9b6504c

  Copy full SHA for 9b6504c

Commits on Sep 16, 2025

• [Docs] Update instructions for how to using existing torch binary (#24892)

  Show description for e1279ef
  

  zou3519

  and

  hmellor

  authored
  e1279ef

  Copy full SHA for e1279ef

Commits on Aug 1, 2025

• [BugFix] Update AttnFusionPass cache key (#21947)

  Show description for 8026a33
  zou3519

  authored
  8026a33

  Copy full SHA for 8026a33

Commits on Jul 29, 2025

• [Bugfix] VLLM_V1 supports passing other compilation levels (#19340)

  Show description for 04e3850
  zou3519

  authored
  04e3850

  Copy full SHA for 04e3850

Commits on Jul 18, 2025

• [Kernel] Apply torch.Tag.needs_fixed_stride_order only for torch==2.6.0 (#19346)

  Show description for b2eb2b5
  zou3519

  authored
  b2eb2b5

  Copy full SHA for b2eb2b5

• [CI] Update CODEOWNERS for vllm/compilation (#21185)

  Show description for 21274ab
  zou3519

  authored
  21274ab

  Copy full SHA for 21274ab

Commits on Jul 15, 2025

• [BugFix] VLLM_DISABLE_COMPILE_CACHE=1 should disable all reads and writes from the cache (#20942)

  Show description for ba8c300
  zou3519

  authored
  ba8c300

  Copy full SHA for ba8c300

Commits on Jul 14, 2025

• [CI] cc folks on changes to vllm/compilation (#20925)

  Show description for 6676246
  zou3519

  authored
  6676246

  Copy full SHA for 6676246

Commits on Jul 12, 2025

• [Bugfix] Fix torch.compile x LoRA for PyTorch 2.8 (#20823)

  Show description for a3a5a47
  zou3519

  authored
  a3a5a47

  Copy full SHA for a3a5a47

Commits on Jun 19, 2025

• [BugFix] Fix use_cudagraph=False (#19612)

  Show description for ed33349
  zou3519

  authored
  ed33349

  Copy full SHA for ed33349

Commits on Jun 15, 2025

• [CI/Build] Fix torch nightly CI dependencies part 2 (#19589)

  zou3519

  authored
  91b2c17

  Copy full SHA for 91b2c17

Commits on Jun 11, 2025

• [CI/Build] Fix torch nightly CI dependencies (#19505)

  Show description for 42f52cc
  zou3519

  authored
  42f52cc

  Copy full SHA for 42f52cc

Commits on Jun 10, 2025

• [BugFix] Allow use_cudagraph to work with dynamic VLLM_USE_V1 (#19390)

  Show description for 77f0d46
  zou3519

  authored
  77f0d46

  Copy full SHA for 77f0d46

Commits on Jun 9, 2025

• [Misc] Cleanup compilation tests (#19343)

  Show description for 3a4d417
  zou3519

  authored
  3a4d417

  Copy full SHA for 3a4d417

Commits on Jun 8, 2025

• [Misc] Change tests/compile to use VLLM_V1 by default (#19302)

  Show description for 3d64d36
  zou3519

  authored
  3d64d36

  Copy full SHA for 3d64d36

• [Bugfix] Re-enable use_cudagraph in vLLM v1 (#19299)

  Show description for eaa2e51
  zou3519

  authored
  eaa2e51

  Copy full SHA for eaa2e51

Commits on Jun 6, 2025

• Fix CompilationConfig repr (#19091)

  Show description for da511d5
  zou3519

  authored
  da511d5

  Copy full SHA for da511d5

Commits on May 30, 2025

• Improve "failed to get the hash of the compiled graph" error (#18956)

  Show description for 84ec470
  zou3519

  authored
  84ec470

  Copy full SHA for 84ec470

Commits on May 29, 2025

• Use standalone_compile by default in torch >= 2.8.0 (#18846)

  Show description for a521ef0
  zou3519

  authored
  a521ef0

  Copy full SHA for a521ef0

Pagination

Previous

Next