Search results

Upstream 091 eplb dynamic

documentationImprovements or additions to documentationImprovements or additions to documentation

merge-conflicts

module:core

module:ops

module:quantization

module:tests

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1663

#1663 In vllm-project/vllm-ascend;

· shiyuan680 opened on Jul 8, 2025

New 091

ci/build

documentationImprovements or additions to documentationImprovements or additions to documentation

merge-conflicts

module:core

module:ops

module:quantization

module:tests

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1658

#1658 In vllm-project/vllm-ascend;

· shiyuan680 opened on Jul 7, 2025

[CustomOP][Refactor] Register CustomOP instead of overwrite forward_oot

module:ops

module:quantization

Status: Draft (not ready).

vllm-project/vllm-ascendnumber 1647

#1647 In vllm-project/vllm-ascend;

· MengqingCao opened on Jul 7, 2025

add mc2 mask

module:ops

module:quantization

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1642

#1642 In vllm-project/vllm-ascend;

· weiguihua2 opened on Jul 7, 2025

Enable the super kernel feature under the Multistream Moe feature

module:core

module:ops

module:quantization

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1641

#1641 In vllm-project/vllm-ascend;

· NNUCJ opened on Jul 7, 2025

pangumoe support rope

module:ops

module:tests

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1633

#1633 In vllm-project/vllm-ascend;

· GDzhu01 opened on Jul 5, 2025

[Refactor] Use tuple as kv cache instead of tensor

merge-conflicts

module:ops

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1594

#1594 In vllm-project/vllm-ascend;

· lidenghui1110 opened on Jul 2, 2025

[Feature]Moe alltoallv communication optimization for unquantized RL training sence & alltoallv support dpo

module:core

module:ops

module:tests

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1547

#1547 In vllm-project/vllm-ascend;

· weijinqian0 opened on Jul 1, 2025

[V0.9.1] Use AddRmsNormQuant ops in the custom model to optimize Qwen3's performance

module:ops

module:quantization

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1545

#1545 In vllm-project/vllm-ascend;

· rjg-lyh opened on Jul 1, 2025

[cherry-pick] Backport multistream MLA fixes and TP communication optimizations

module:ops

module:tests

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1474

#1474 In vllm-project/vllm-ascend;

· sdmyzlp opened on Jun 27, 2025

add qwen3-moe optimization

merge-conflicts

module:ops

module:tests

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1441

#1441 In vllm-project/vllm-ascend;

· shiyuan680 opened on Jun 26, 2025

rm router logits Improve TTOP 3ms

module:ops

Status: Open (in progress).

vllm-project/vllm-ascendnumber 1407

#1407 In vllm-project/vllm-ascend;

· ttanzhiqiang opened on Jun 24, 2025