skills.rest
Fudan-SMI-lab
Fudan-SMI-labOFFICIAL
Explore all skills by this author →
GitHub

e2e-v2-0dc9b40e0da8-promoted

Create virtual environments inheriting system torch-npu stacks on Ascend hosts.

39|4|Updated May 18, 2026
Search Tags:#venv#environment-setup#ascend-npu#system-site-packages#cuda-variant-avoidance#ml-migration
One-click install
npx skills add https://github.com/Fudan-SMI-lab/SEAM --skill e2e-v2-0dc9b40e0da8-promoted
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: e2e-v2-0dc9b40e0da8-promoted
Source: https://github.com/Fudan-SMI-lab/SEAM/tree/main/.memory/skills/e2e-v2-0dc9b40e0da8-promoted
Command: npx skills add https://github.com/Fudan-SMI-lab/SEAM --skill e2e-v2-0dc9b40e0da8-promoted

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

System-site-packages venv strategy eliminates the need to download CUDA variant wheels on pre-configured Ascend hosts by leveraging the system Python's torch-npu stack.

Core Features & Use Cases

  • Inherit system-wide torch-npu and related stacks into a new virtual environment to avoid large wheel downloads.
  • Speed up environment setup on Ascend-based deployments and reduce network dependency during migrations.
  • Use Case: When migrating ML projects to Ascend hardware with pre-installed stacks, this approach minimizes setup time and ensures compatibility.

Quick Start

Create a virtual environment with --system-site-packages and verify torch.npu is available.

Frequently Asked Questions about e2e-v2-0dc9b40e0da8-promoted

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I avoid downloading CUDA variant wheels when migrating ML projects to Ascend hardware?▼

A system-site-packages venv strategy prevents CUDA variant downloads by inheriting the pre-configured system Python's torch-npu stack. It leverages existing Ascend hardware packages to enable fast, reproducible ML migrations with minimal additional pip installs.

How to set up a Python venv on Ascend NPU without reinstalling torch-npu?▼

Create the virtual environment with --system-site-packages enabled to directly inherit the system-wide torch-npu and related stacks. This requires minimal additional pip installs and avoids redundant package downloads during ML migration.

Can I use --system-site-packages to speed up environment setup on pre-configured Ascend hosts?▼

Yes, using --system-site-packages speeds up environment setup on pre-configured Ascend hosts by inheriting the existing system Python torch-npu stack. This reduces network dependency and enables fast, reproducible ML project migrations.

What is the best way to validate torch.npu availability in a new virtual environment?▼

The best way to validate torch.npu availability is by running a simple availability check after creating the venv. This verifies that the system-site-packages strategy successfully inherited the Ascend torch-npu stack into the new environment.

When should I not use the system-site-packages venv strategy for Ascend migrations?▼

Avoid the system-site-packages venv strategy if your target environment lacks a pre-configured torch-npu stack. This approach applies specifically when the system ships the required packages and relies on inheriting them to avoid CUDA variant downloads.

Related Skills

Explore More Software Engineering Skills→
Fudan-SMI-labFudan-SMI-lab
39

torch-npu-venv-setup-cpu-base

Pin CPU-base PyTorch and install matching torch-npu and torchvision.

Official
Intermediate
Fudan-SMI-labFudan-SMI-lab
39

torch-npu-pyyaml-preinstall

Preinstall PyYAML before torch-npu to prevent transitive dependency failures.

Official
Intermediate
FeRhodiumFeRhodium
7

dependency-analysis

Analyze dependency files to identify CUDA packages and assess Ascend NPU compatibility.

Community
Intermediate
sayalinvidiasayalinvidia

cupynumeric-install

Install cuPyNumeric via conda or pip and verify in an isolated environment.

Community
Advanced