BBuf/vllm-gemma4-optimization
skills/model-optimization/vllm/vllm-gemma4-optimization/SKILL.md
PR-backed optimization manual for Gemma 4 in vLLM. Use when an engineer needs to audit, debug, extend, or document Gemma 4 text, MoE, multimodal, reasoning, tool use, and quantized MoE serving.
$ install --global
skillsauthnpx skillsauth add BBuf/AI-Infra-Auto-Driven-SKILLS vllm-gemma4-optimizationInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: May 2, 2026, 2:38 AM136.8s2 files scanned
SKILL.md
- name:
- vllm-gemma4-optimization
- description:
- PR-backed optimization manual for Gemma 4 in vLLM. Use when an engineer needs to audit, debug, extend, or document Gemma 4 text, MoE, multimodal, reasoning, tool use, and quantized MoE serving.
vLLM Gemma 4 Optimization
Overview
This skill covers Gemma 4 text, MoE, multimodal, reasoning, tool use, and quantized MoE serving.
Evidence snapshot:
- vLLM mainline checked around
0f7be0f2f76814f80f9091220a5fbbb53912ad00 - Canonical PR notes:
references/pr-history.md - History mirrors:
model-pr-optimization-history/vllm/gemma4/README.zh.mdandREADME.en.md
Non-Negotiable Evidence Rule
Use skills/model-optimization/model-pr-diff-dossier/SKILL.md as the production bar.
Every PR cited for this family must be based on diff reading, not only PR titles.
Runtime Surfaces
vllm/vllm/model_executor/models/gemma4.pyvllm/vllm/model_executor/models/gemma4_mm.py
Current Main Summary
- Gemma 4 is a genuinely multi-surface family: text, MoE, multimodal, tool calling, and speculative decoding all matter.
- Fast prefill, quantized MoE, and tool-parser correctness are the main areas that changed rapidly after bring-up.
Key Landed PRs
- #38826
Implement Google Gemma 4 architecture support: Initial Gemma 4 text/MoE/multimodal landing. - #38879
Enable Fast Prefill Optimization: Added YOCO KV-sharing based fast prefill for Gemma4. - #39045
Support quantized MoE: Extended Gemma4 to quantized MoE checkpoints. - #38844
Enable Gemma4ForCausalLM to load LoRA adapters correctly: Fixed adapter naming/load behavior. - #39450
Add Gemma4 Eagle3 support: Enabled speculative decode for Gemma4.
Validation Lanes
- Startup on the current vLLM mainline checkpoint.
- Re-run the parser, quantization, MTP, or multimodal lane that matches this family.
- Re-check tests after touching loader or processor code.
Related Skills
BBuf/vllm-sota-humanize-loop
development
VerifiedTrustedCommunity
Run an autonomous Humanize-governed vLLM SOTA performance loop for one LLM model: first perform the fixed fair vLLM/SGLang/TensorRT-LLM deployment search and benchmark, then start one RLCR loop that repeatedly decides the gap, profiles the current bottleneck, runs layer/kernel pipeline analysis, patches vLLM code, optionally uses ncu-report-skill for kernel evidence, and revalidates until vLLM matches or beats the best observed framework under the same workload and SLA.
423SKILL.mdUpdated May 27, 2026
BBuf/llm-pipeline-analysis
devops
VerifiedTrustedCommunity
Inspect LLM torch profiler traces at forward-pass, layer, and kernel level. Use when you need layer timings, anchor-kernel boundaries, representative kernel flows, or Perfetto time ranges.
423SKILL.mdUpdated May 21, 2026
BBuf/sglang-sota-humanize-loop
development
VerifiedTrustedCommunity
Run an autonomous Humanize-governed SGLang SOTA performance loop for one LLM model: first perform the fixed fair SGLang/vLLM/TensorRT-LLM deployment search and benchmark, then start one RLCR loop that repeatedly decides the gap, profiles the current bottleneck, runs layer/kernel pipeline analysis, patches SGLang code, optionally uses ncu-report-skill for kernel evidence, and revalidates until SGLang matches or beats the best observed framework under the same workload and SLA.
423SKILL.mdUpdated May 17, 2026
BBuf/model-pr-history-knowledge
documentation
VerifiedTrustedCommunity
Use when an SGLang, vLLM, or TensorRT-LLM serving/model optimization task needs prior model-family PR evidence. Query and read the PR-driven history docs under model-pr-optimization-history before choosing source paths, fast paths, kernel/fusion ideas, regression risks, or validation lanes.
423SKILL.mdUpdated May 17, 2026