Skills
.Work. You
Rest

vllm-glm5-glm51-optimization

Community

PR-backed GLM-5 optimization for vLLM.

Software Engineering#optimization#deepseek#vllm#glm-5#glm-5.1#causal-lm
AuthorBBuf
Version1.0.0
Installs0

System Documentation

What problem does it solve?

GLM-5 integration in vLLM often requires careful PR-based tracking and adaptation. This manual provides a PR-backed optimization approach to audit, debug, extend, or document the current partial GLM-5 bring-up, including aliasing into the DeepSeek-V2/V3 runtime, rope interleave handling, and MTP correctness.

Core Features & Use Cases

  • PR-diff driven optimization guidance for GLM-5/5.1 in vLLM.
  • Guidance on integrating GlmMoeDsaForCausalLM aliasing into DeepSeek-V2/V3.
  • Notes on rope interleave handling and MTP logit stability.
  • References to PR history cards for audit trail.

Quick Start

Audit the current GLM-5 bring-up in vLLM using the PR-backed dossier and implement the recommended aliasing and MTP adjustments.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: vllm-glm5-glm51-optimization
Download link: https://github.com/BBuf/AI-Infra-Auto-Driven-SKILLS/archive/main.zip#vllm-glm5-glm51-optimization

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.