[cherry-pick][0.9.1] rebase main#1250

Merged

ganyi1996ppo merged 28 commits into

vllm-project:v0.9.1-devfrom

momo609:091dev

Jun 17, 2025

Commits on Jun 17, 2025

[CI] remove old quantization model (#1003)

22dimensions
authored and
wangxiaoxin (A)
committed
Update 0.9.0rc1 contributors info (#1148)

Yikun
authored and
wangxiaoxin (A)
committed
[CI] Make accuarcy CI and report work (#1078)

zhangxinyuehfad
authored and
wangxiaoxin (A)
committed
[Bugfix] add compilation/__init__.py to fix import error (#1152)

wangxiyuan
authored and
wangxiaoxin (A)
committed
[CI] Run e2e after pre check pass (#1132)

wangxiyuan
authored and
wangxiaoxin (A)
committed
[MLA][Graph] Improve assertion on Graph mode with MLA (#933)

MengqingCao
authored and
wangxiaoxin (A)
committed
[CI] rename Qwen2.5-0.5B-Instruct-W8A8 model (#1145)

22dimensions
authored and
wangxiaoxin (A)
committed
[CI] Skip test_v1_spec_decode.py::test_ngram_correctness to make longterm CI pass (#1163)

MengqingCao
authored and
wangxiaoxin (A)
committed
Support multistream of shared experts in FusedMoE (#997)

sdmyzlp
authored and
wangxiaoxin (A)
committed
provide an e2e guide for execute duration profiling (#1113)

depeng1994
authored and
wangxiaoxin (A)
committed
etp best a2 (#1101)

ttanzhiqiang
authored and
wangxiaoxin (A)
committed
[Doc] Fix the config parameter name "enable" in graph_mode.md. (#1159)

yzim
authored and
wangxiaoxin (A)
committed
Enable kvcache_nz for the decode process in torchair graph mode (#1098)

chenwaner
authored and
wangxiaoxin (A)
committed
[CI] Upgrade vllm to 0.9.1 (#1165)

wangxiyuan
authored and
wangxiaoxin (A)
committed
[Scheduler][MTP] Add support for speculative decoding in AsecendScheduler. (#943)

whx-sjtu
authored and
wangxiaoxin (A)
committed
add custom ascendc kernel vocabparallelembedding (#796)

ttanzhiqiang
authored and
wangxiaoxin (A)
committed
[CI][Benchmark] Add new model and v1 test to perf benchmarks (#1099)

Potabk
authored and
wangxiaoxin (A)
committed
[CI][Benchmark] Add qwen2.5-7b test (#1104)

Potabk
authored and
wangxiaoxin (A)
committed
[fix] fix bug in 1p1d disaggregated_prefill example (#1184)

wangyanhui-cmss
authored and
wangxiaoxin (A)
committed
[Doc] Add Referer header for CANN package download url. (#1192)

wonderful199082
authored and
wangxiaoxin (A)
committed
Support multistream of MLA vector operations (#1135)

sdmyzlp
authored and
wangxiaoxin (A)
committed
[CI] Recover ut for ascend scheduler only in ci of v1. (#1180)

whx-sjtu
authored and
wangxiaoxin (A)
committed
Add ShouJian Zheng (@jianzs) as vLLM Ascend maintainer (#1203)

Yikun
authored and
wangxiaoxin (A)
committed
[CI/UT][Graph] Add ut for torchair graph mode (#1103)

MengqingCao
authored and
wangxiaoxin (A)
committed
[Doc] fix VLLM_USE_V1 value in graph mode docs (#1226)

22dimensions
authored and
wangxiaoxin (A)
committed
Fix the device error when using ray as vllm-acend backend (#884)

zhuo97
authored and
wangxiaoxin (A)
committed
remove main vll verison.
wangxiaoxin (A)
committed
Revert "Add ShouJian Zheng (@jianzs) as vLLM Ascend maintainer (#1203)"
wangxiaoxin (A)
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[cherry-pick][0.9.1] rebase main#1250

[cherry-pick][0.9.1] rebase main#1250
ganyi1996ppo merged 28 commits into
vllm-project:v0.9.1-devfrom
momo609:091dev

Commits on Jun 17, 2025

[CI] remove old quantization model (#1003)

Update 0.9.0rc1 contributors info (#1148)

[CI] Make accuarcy CI and report work (#1078)

[Bugfix] add compilation/init.py to fix import error (#1152)

[CI] Run e2e after pre check pass (#1132)

[MLA][Graph] Improve assertion on Graph mode with MLA (#933)

[CI] rename Qwen2.5-0.5B-Instruct-W8A8 model (#1145)

[CI] Skip test_v1_spec_decode.py::test_ngram_correctness to make longterm CI pass (#1163)

Support multistream of shared experts in FusedMoE (#997)

provide an e2e guide for execute duration profiling (#1113)

etp best a2 (#1101)

[Doc] Fix the config parameter name "enable" in graph_mode.md. (#1159)

Enable kvcache_nz for the decode process in torchair graph mode (#1098)

[CI] Upgrade vllm to 0.9.1 (#1165)

[Scheduler][MTP] Add support for speculative decoding in AsecendScheduler. (#943)

add custom ascendc kernel vocabparallelembedding (#796)

[CI][Benchmark] Add new model and v1 test to perf benchmarks (#1099)

[CI][Benchmark] Add qwen2.5-7b test (#1104)

[fix] fix bug in 1p1d disaggregated_prefill example (#1184)

[Doc] Add Referer header for CANN package download url. (#1192)

Support multistream of MLA vector operations (#1135)

[CI] Recover ut for ascend scheduler only in ci of v1. (#1180)

Add ShouJian Zheng (@jianzs) as vLLM Ascend maintainer (#1203)

[CI/UT][Graph] Add ut for torchair graph mode (#1103)

[Doc] fix VLLM_USE_V1 value in graph mode docs (#1226)

Fix the device error when using ray as vllm-acend backend (#884)

remove main vll verison.

Revert "Add ShouJian Zheng (@jianzs) as vLLM Ascend maintainer (#1203)"