GitLab

修复awq的triton支持，和moe模型的接口bug，以及awq_moe_marlin的接口相关问题，以及解决一些w4a16的精度问题

See merge request dcutoolkit/deeplearing/vllm!403

# Conflicts:
#   vllm/model_executor/layers/fused_moe/layer.py

[feat]1.适配ds w8a8及mtp;2.添加宽松mtp;3.适配w8a8 DEEPEP;4.解决ds 671B精度异常

See merge request dcutoolkit/deeplearing/vllm!400

修复vit attn的导入问题，以及w4a16的gptq的接口问题

See merge request dcutoolkit/deeplearing/vllm!395

修复了awq的shape的bug，以及兼容了lmslim注册导入的的情况

See merge request dcutoolkit/deeplearing/vllm!394

Signed-off-by: Shengqi Chen <harry-chen@outlook.com>
(cherry picked from commit 136c499f)

Signed-off-by: NickLucche <nlucches@redhat.com>
(cherry picked from commit ea6102b8)

Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
(cherry picked from commit 444e2e7e)

Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
(cherry picked from commit 8ebf271b)

Signed-off-by: Shengqi Chen <harry-chen@outlook.com>
(cherry picked from commit 965765ae)

[CI] Implement uploading to PyPI and GitHub in the release pipeline, enable release image building for CUDA 13.0 (#31032)

(cherry picked from commit 8e61425e)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

区分pcie和hglink custom allreduce的使用
vllm：export VLLM_CUSTOM_CACHE=1
dtk：export HIP_KERNEL_EVENT_SYSTENFENCE=1

set VLLM_USE_FUSED_RMS_ROPE=1
add SUPPORT_MOE_MARLIN_W16A16 to use moe marlin on bw
support fa kvcache fp8 (todo: add VLLM_USE_QUERY_QUANT to not use q quant)
update moe_align_block_size

fix _forward_encoder_attention
remove medusa
set VLLM_PCIE_USE_CUSTOM_ALLREDUCE=1

Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>

Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
(cherry picked from commit bcf2333c)

Signed-off-by: Kevin H. Luu <khluu000@gmail.com>

Signed-off-by: mgoin <mgoin64@gmail.com>
(cherry picked from commit 1be5a735)

Signed-off-by: ganyi <ygan@amd.com>
(cherry picked from commit 77c16df3)

Signed-off-by: Doug Lehr <douglehr@amd.com>
Co-authored-by: Doug Lehr <douglehr@amd.com>
(cherry picked from commit c5891b54)

[Bugfix][ROCm][performance] Resolve the performance regression issue of the Qwen3-Next-80B-A3B-Thinking under rocm_atten (#32336)
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
(cherry picked from commit e27078ea)