-
-
Notifications
You must be signed in to change notification settings - Fork 9.4k
Pull requests: vllm-project/vllm
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[misc] ignore marlin_moe_wna16 local gen codes
ready
ONLY add when PR is ready to merge/full CI is needed
#16760
by DefTruth
was merged Apr 17, 2025
Loading…
[Bugfix][Kernel] fix potential cuda graph broken for merge_attn_states kernel
ready
ONLY add when PR is ready to merge/full CI is needed
#16693
by DefTruth
was merged Apr 16, 2025
Loading…
[Misc] remove warning if triton>=3.2.0
ready
ONLY add when PR is ready to merge/full CI is needed
#16553
by DefTruth
was merged Apr 14, 2025
Loading…
Revert "[Model] use AutoWeightsLoader for deepseek_v2, internlm2"
ready
ONLY add when PR is ready to merge/full CI is needed
#16453
by DefTruth
was merged Apr 11, 2025
Loading…
[Kernel] Remove redundant Exp calculations
ready
ONLY add when PR is ready to merge/full CI is needed
#16123
by DefTruth
was merged Apr 15, 2025
Loading…
[Bugfix] hotfix for gptq-marlin non-contiguous error
#15374
by DefTruth
was closed Mar 25, 2025
Loading…
[Misc] Add tuned R1 w8a8 and MoE configs for NVIDIA L20
ready
ONLY add when PR is ready to merge/full CI is needed
#15322
by DefTruth
was merged Mar 23, 2025
Loading…
[Bugfix] fix torch.compiled cache hash error
ready
ONLY add when PR is ready to merge/full CI is needed
#14953
by DefTruth
was merged Mar 23, 2025
Loading…
[Bugfix][V1] Fix compiled graph hash
ci/build
documentation
Improvements or additions to documentation
needs-rebase
v1
#14867
by DefTruth
was closed Mar 17, 2025
Loading…
[VLM][Bugfix] enable internvl running with num_scheduler_steps > 1
ready
ONLY add when PR is ready to merge/full CI is needed
#8614
by DefTruth
was merged Sep 25, 2024
Loading…
[Bugfix][Kernel] allow non-power-of-2 for prefix prefill with alibi
#4573
by DefTruth
was merged May 8, 2024
Loading…
[Bugfix][Kernel] allow non-power-of-2 head sizes for prefix prefill with alibi
#4489
by DefTruth
was closed May 3, 2024
Loading…
Previous Next
ProTip!
Exclude everything labeled
bug
with -label:bug.