[DSV4] Cherry pick missing commits from deepseek_v4 branch and enhance tests (#24793)

Co-authored-by: Xinyuan Tong <115166877+JustinTong0323@users.noreply.github.com>
Co-authored-by: yueming-yuan <yym022502@gmail.com>
This commit is contained in:
Baizhou Zhang
2026-05-09 04:15:37 -07:00
committed by GitHub
co-authored by Xinyuan Tong yueming-yuan
parent 4b23f6bdc5
commit ef5e9f8aba
15 changed files with 481 additions and 87 deletions
@@ -378,7 +378,6 @@ export const DeepSeekV4Deployment = () => {
"SGLANG_OPT_USE_CUSTOM_ALL_REDUCE_V2=1",
"SGLANG_OPT_SWA_EVICT_DROP_PAGE_MARGIN=1",
"SGLANG_OPT_USE_DEEPGEMM_MEGA_MOE=0",
"SGLANG_OPT_FIX_HASH_MEGA_MOE=0",
"SGLANG_OPT_USE_FAST_MASK_EP=1",
"SGLANG_OPT_FIX_MEGA_MOE_MEMORY=1",
"SGLANG_OPT_DEEPGEMM_MEGA_MOE_NUM_MAX_TOKENS_PER_RANK=4096",
@@ -412,7 +411,6 @@ export const DeepSeekV4Deployment = () => {
"NVSHMEM_DISABLE_IB=1",
"SGLANG_OPT_SWA_RELEASE_LEAF_LOCK_AFTER_WINDOW=1",
"SGLANG_OPT_USE_DEEPGEMM_MEGA_MOE=1",
"SGLANG_OPT_FIX_HASH_MEGA_MOE=1",
"SGLANG_OPT_DEEPGEMM_MEGA_MOE_NUM_MAX_TOKENS_PER_RANK=8320",
);
} else {
@@ -926,7 +924,6 @@ python3 -m sglang_router.launch_router \\
# And set these env vars:
SGLANG_OPT_USE_DEEPGEMM_MEGA_MOE=1
SGLANG_OPT_FIX_HASH_MEGA_MOE=1
SGLANG_OPT_FIX_MEGA_MOE_MEMORY=1
SGLANG_OPT_FIX_NEXTN_MEGA_MOE=1
SGLANG_OPT_DEEPGEMM_MEGA_MOE_NUM_MAX_TOKENS_PER_RANK=8320