Updated on 2026-06-01 GMT+08:00

ModelArts Best Practices

This document provides ModelArts samples concerning a variety of scenarios and AI engines to help you quickly understand the process and operations of using ModelArts for AI development.

All practical case studies provided in this document are intended solely for scenarios such as function validation and prototype testing. Given the stringent requirements of production environments for high availability, security compliance, observability, and operations management, do not use the case studies in this document directly in production environments.

LLM Training and Inference

Sample	Scenario	Description
Adapting Mainstream Open-Source Models for NPU Inference on Ascend-vLLM with PyTorch	Inference deployment, inference performance test, inference accuracy test, and inference model quantization	Describes the inference deployment process of mainstream open-source models, such as Llama, Qwen3, and Qwen-VL series, based on ModelArts. The Ascend-vLLM framework and Ascend NPUs are used for inference.

Image Generation Model Training and Inference

Sample	Scenario	Description
Adapting Stable Diffusion for NPU Inference with Diffusers/ComfyUI and Lite Server (6.5.907)	SD1.5, SDXL, SD3.5, and Hunyuan model inference	Describes the inference process of mainstream image generation models based on ModelArts Lite Server. Ascend NPUs are used for inference. After the inference service is started, it can be used in image generation scenarios.
Stable Diffusion XL Inference Guide Based on ModelArts Notebook (6.5.907)	SDXL model inference	Describes the inference process of mainstream image generation models based on Notebook. Ascend NPUs are used for inference. After the inference service is started, it can be used in image generation scenarios.

Sample

Scenario

Description

Adapting Stable Diffusion for NPU Inference with Diffusers/ComfyUI and Lite Server (6.5.907)

SD1.5, SDXL, SD3.5, and Hunyuan model inference

Describes the inference process of mainstream image generation models based on ModelArts Lite Server. Ascend NPUs are used for inference.

After the inference service is started, it can be used in image generation scenarios.

Stable Diffusion XL Inference Guide Based on ModelArts Notebook (6.5.907)

SDXL model inference

Describes the inference process of mainstream image generation models based on Notebook. Ascend NPUs are used for inference.

After the inference service is started, it can be used in image generation scenarios.

Video Generation Model Training and Inference

Sample	Scenario	Description
Inference Guide for Wan2.1, Wan2.2, HunyuanVideo, and CogVideo Series Models Adapted to NPU via ModelArts Lite Server	Wan series model inference	Describes the inference process of Wan series models based on ModelArts Lite Server. The PyTorch framework and Ascend NPUs are used for inference.

ModelArts Notebook Cases

**Table 1** ModelArts samples
Sample	Function	Scenario	Description
Migrating the Conda Environment on a Notebook Instance to an SFS Disk	Environment migration	Development environments	Describes how to migrate the Conda environment of a notebook instance to an SFS disk.

ModelArts Model Training Cases

**Table 2** Custom algorithm samples
Sample	Image	Function	Scenario	Description
Building a Handwritten Digit Recognition Model with ModelArts Standard	PyTorch	Algorithm customization	Handwritten digit recognition	Use your customized algorithm to train a handwritten digit recognition model and deploy the model for prediction.

ModelArts Inference Deployment

**Table 3** Inference deployment samples
Sample	Scenario	Description
Migrating a Third-Party Inference Framework to a Custom Inference Engine	Third-party frameworks Inference deployment	ModelArts allows the deployment of third-party inference frameworks. This section describes how to migrate TF Serving and Triton to a custom inference engine.