Install and verify ONNX Runtime stacks for AMD Quark workflows.
npx skills add https://github.com/amd/Quark --skill quark-onnx-install
Please help me install this Agent Skill. Skill: quark-onnx-install Source: https://github.com/amd/Quark/tree/main/.claude/skills-impl/l1-atomic/onnx/quark-onnx-install Command: npx skills add https://github.com/amd/Quark --skill quark-onnx-install
This Skill removes the guesswork from setting up ONNX Runtime for AMD Quark’s ONNX-to-ONNX workflow, helping you avoid provider mismatches, missing execution providers, and import failures before quantization begins.
import onnxruntime fails or the wrong execution provider is available, this Skill guides the exact reinstall and verification path.Ask the skill to diagnose your accelerator backend and produce the exact ONNX Runtime installation and verification steps for Quark.
High-intent search queries and answers about installing and using this skill.
To fix ONNX Runtime import failures, diagnose your accelerator backend and reinstall the correct runtime variant for your CPU, CUDA, or ROCm environment. This ensures version compatibility and resolves execution provider mismatches before quantization begins.
The wrong execution provider appears due to package conflicts or version mismatches between ONNX Runtime and your CUDA or ROCm environment. Version-aware package selection and provider validation are required to align the runtime with your specific backend.
The best way to verify ONNX Runtime compatibility is to check ONNX, ONNX Runtime, and related package versions against Quark’s supported ranges. This compatibility verification prevents import failures and custom-ops compilation issues during quantization.
Yes, you need to ensure Quark’s ONNX custom operators can compile and load correctly in your local environment. Custom ops readiness checks validate that the installed runtime stack supports the specific operations required for quantization.
To install ONNX Runtime for ROCm environments, apply backend-specific installation planning to choose the correct ROCm-compatible runtime variant. This approach avoids execution provider mismatches and package conflicts that block AMD Quark quantization.
amdInstalls AMD Quark and verifies its Python package environment after PyTorch setup.
amdExecute end-to-end post-training quantization workflows for ONNX models with AMD Quark.
amdDiagnose Quark ONNX installation, calibration, quantization, and custom-op failures.
amdGuide end-to-end post-training quantization of ONNX models with AMD Quark.