ascend-ai-coding/awesome-ascend-skills

14 resources in this repository

🎯14Skills

🎯Skills14

A complete toolkit for Huawei Ascend NPU model conversion and inference, supporting ONNX to .om format conversion via ATC, Python inference with ais_bench, precision comparison, and end-to-end YOLO inference.

atc-model-converter

🎯npu-smi🎯Skill

A command reference skill for Huawei Ascend NPU npu-smi tool, covering device queries (health, temperature, power, memory), configuration, firmware upgrades, virtualization (vNPU), and certificate management.

npu-smi

🎯hccl-test🎯Skill

HCCL (Huawei Collective Communication Library) performance testing skill for Ascend NPU clusters, covering distributed communication bandwidth testing, HCCL functionality verification, and collective operations benchmarking.

hccl-test

🎯ascend-docker🎯Skill

Streamlines Docker container creation, optimization, and deployment for AI/ML workloads on Ascend hardware with performance-focused configurations.

ascend-docker

🎯msmodelslim🎯Skill

Optimizes machine learning model size and inference speed through advanced pruning, quantization, and compression techniques for efficient deployment.

msmodelslim

🎯ais-bench🎯Skill

Benchmarks and profiles AI model performance across hardware, measuring inference speed, memory usage, and computational efficiency.

Skill

Skill

Skill