昇腾开源
🏁 开始使用
快速安装昇腾环境
🏗️ 基础设施与框架
accelerate
DeepSpeed
kernels
pytorch
Ray
transformers
🧠 训练与微调框架
LLaMA-Factory
ms-swift
ROLL
torchtitan
Transformer Reinforcement Learning
VeOmni
Installation with Ascend NPU (ARM)
Installation with Ascend NPU (x86)
Get Started with Ascend NPU
verl
PyG
🚀 推理与服务
llama.cpp
lmdeploy
onnxruntime
sentence-transformers
sglang
torchchat
vLLM-Ascend
⚙️ 算子开发与编程
Triton-Ascend
🎨 多模态、应用与评测
diffusers
lm-evaluation-harness
open_clip
opencompass
opencv
stable-diffusion-webui
pytorch-image-models
wenet
whisper.cpp
昇腾开源
VeOmni
查看页面源码
VeOmni
Installation with Ascend NPU (ARM)
Required Environment
Prepare CANN
Install with pip
Set up CANN environment before installing torchcodec
Video/Audio Processing Dependencies (Optional)
Installation with Ascend NPU (x86)
Required Environment
Prepare CANN
Install with uv or pip
UV
Pip
Set up CANN environment before installing torchcodec
Video/Audio Processing Dependencies (Optional)
Get Started with Ascend NPU
Overview
What This Guide Covers
Key Updates
Installation
Installation Options
Detailed Installation Guide
Docker Support
Version Compatibility
Supported Models
Typical Usage
Common Precision Issues and Solutions
Ascend Profiling Collection and Analysis
FAQ
Declarations