graph-mode-adaptation

Community

Accelerate NPU graph-mode inference.

Authortuliang1024
Version1.0.0
Installs0

System Documentation

What problem does it solve?

图模式适配技能帮助开发者将模型在 NPU 平台通过 torch.compile 的图模式进行加速,同时提供设计、调试与验证的系统性工作流,解决图中断、动态形状、后端差异等常见挑战。

Core Features & Use Cases

  • 提供 npugraph_ex 与 GE 两种图模式的适配指南与对比分析,帮助选择合适的后端。
  • 指导识别与修复 Graph Break、重编译及动态形状带来的问题,确保稳定性与性能。
  • 给出端到端的工作流模板,包括方案设计、方案确认、实现与验证,以及对 LLM 与通用模型的要点区分。

Quick Start

Describe a ready-to-implement plan to adapt a model to NPU graph mode with torch.compile and verify its correctness and performance.

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: graph-mode-adaptation
Download link: https://github.com/tuliang1024/cann-recipes-infer/archive/main.zip#graph-mode-adaptation

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.