Skip to content

[Klaud Cold] Update qwen3.8next-fp4-b300-sglang-agentic-mtp SGLang image to dev-cu13-qwen38-next-local (2026-09-07, digest-pinned) #2879

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
functionstackx wants to merge 2 commits into
base: main
Choose a base branch
from
+11 −2
Open

[Klaud Cold] Update qwen3.8next-fp4-b300-sglang-agentic-mtp SGLang image to dev-cu13-qwen38-next-local (2026-09-07, digest-pinned) #2879

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c2fc75c
[Klaud Cold] Update qwen3.8next-fp4-b300-sglang-agentic-mtp SGLang im…
functionstackx Sep 8, 2026
a50e00e
Move the B300 recipe to cluster:b300-dsxe; cluster:b300-nv was retire…
functionstackx Sep 8, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions configs/nvidia-master.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7237,10 +7237,10 @@ qwen3.5-fp4-b300-sglang-agentic-mtp:
# Day-zero recipe; mirrors the B200 arm. TP1 per the cookbook's verified
# single-node command: 126 GiB of NVFP4 weights fit on one B300.
qwen3.8next-fp4-b300-sglang-agentic-mtp:
image: lmsysorg/sglang:qwen38flashnext
image: lmsysorg/sglang:dev-cu13-qwen38-next-local@sha256:9d2a843c706c74bc259c0d9abf360551eb2734e1e7d255ab012a6965f10480b6
model: RadixArk/Qwen3.8-Flash-Next-NVFP4
model-prefix: qwen3.8next
runner: cluster:b300-nv
runner: cluster:b300-dsxe
precision: fp4
framework: sglang
multinode: false
Expand Down
9 changes: 9 additions & 0 deletions perf-changelog.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6921,3 +6921,12 @@
- "Pick up the latest automatic ROCm DeepSeek-V4 optimizations, including fused mHC post/pre plus RMSNorm, gfx950 C4A top-k dispatch, fused C4 compressor GEMMs, fused SWA q/kv RMSNorm plus q FP8 quantization, and medium-batch cooperative top-k tuning."
- "Keep the existing VLLM_ROCM_USE_AITER=1, VLLM_ROCM_USE_AITER_MOE=1, and --moe-backend aiter settings, and explicitly add VLLM_ROCM_USE_AITER_FUSION_SHARED_EXPERTS=1 plus VLLM_ROCM_QUICK_REDUCE_QUANTIZATION=INT4 to both STP and MTP paths. The current checkpoint's shared-expert path does not satisfy the latest vLLM fusion conditions, so that fusion flag self-disables while preserving recipe parity."
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2792

- config-keys:
- qwen3.8next-fp4-b300-sglang-agentic-mtp
scenario-type:
- agentic-coding
description:
- "Move the image from the mutable model-branch tag lmsysorg/sglang:qwen38flashnext (last re-pushed 2026-09-03T20:54:46Z, digest sha256:5ae5816783d58e2e56e84d2e863f5441425056f500b7fbd7448c4aae017a2521) to the newest Qwen3.8-Flash-Next model-branch build, lmsysorg/sglang:dev-cu13-qwen38-next-local@sha256:9d2a843c706c74bc259c0d9abf360551eb2734e1e7d255ab012a6965f10480b6 (Docker Hub last pushed 2026-09-07T11:12:42Z), pinned by digest so every node runs the same build; the same move as the H100/H200 siblings in #2876/#2877. Qwen3.8-Flash-Next (architecture Qwen4ExpForConditionalGeneration) is not in SGLang main, v0.5.19, or the 2026-09-07 nightly (nightly-dev-cu13-20260907-30705c00), so no upstream nightly can serve it; the model-branch dev images are the only lineage. benchmarks/single_node/agentic/qwen3.8next_fp4_b300_sglang_mtp.sh is unchanged, as are the serve flags, golden acceptance length, and concurrency grid. The sweep is the validation that this branch build still carries the fixes the recipe relies on."

@claude claude Bot Sep 8, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🔴 The new entry's pr-link is the literal string "PRLINK_PLACEHOLDER", which is not a value the changelog tooling recognizes: utils/validate_perf_changelog.py's PR_LINK_PLACEHOLDERS only accepts "XXX" or "https://github.com/SemiAnalysisAI/InferenceX/pull/XXX", so validate_added_pr_link raises ChangelogValidationError and the CI gate fails; the merge-prep step (utils/prepare_perf_changelog_merge.py canonicalize_appended_links) also rejects it with "unexpected pr-link" instead of rewriting it to the real PR URL, so this PR cannot pass the changelog gate or be auto-fixed at merge as the repo conventions assume. Fix: use one of the accepted placeholders ("XXX" or the pull/XXX form) for the pr-link field of any newly appended entry.

Extended reasoning...

utils/validate_perf_changelog.py:144-160 validate_added_pr_link(link, pr_number) computes expected = f".../pull/{pr_number}" and raises unless link is in PR_LINK_PLACEHOLDERS ({"XXX", ".../pull/XXX"}) or equals expected. compare_entries (line ~206) calls this for every added entry, and main() is the CI entry point, so any PR run of the gate against this changelog entry (pr-link: PRLINK_PLACEHOLDER) fails validation. Separately, utils/prepare_perf_changelog_merge.py:109-113 canonicalize_appended_links checks if link not in PR_LINK_PLACEHOLDERS: raise ChangelogValidationError(f"appended entry {index+1} has unexpected pr-link {link!r}"), so the merge-with-reuse tooling that is supposed to rewrite placeholders to the real PR URL also fails on this exact string instead of silently fixing it, contradicting the repo's documented merge flow.

Verification: normal. The appended perf-changelog.yaml entry ends with pr-link: PRLINK_PLACEHOLDER. The recognized set at utils/validate_perf_changelog.py:24-27 is PR_LINK_PLACEHOLDERS = {"XXX", "https://github.com/SemiAnalysisAI/InferenceX/pull/XXX"}; "PRLINK_PLACEHOLDER" is not in it. utils/prepare_perf_changelog_merge.py canonicalize_appended_links (lines ~110-114): link =…

- "Move the recipe from the retired cluster:b300-nv fleet (launcher and runner labels removed in #2826) to cluster:b300-dsxe so the sweep has a runner to schedule on; same repoint as #2829 for the GLM-5.2 B300 sibling. The runner change means this is not an append-only bump: the whole curve reruns on DSXE."
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2879
Loading