Execute end-to-end post-training quantization workflows for ONNX models with AMD Quark.
npx skills add https://github.com/amd/Quark --skill quark-onnx-ptq
Please help me install this Agent Skill. Skill: quark-onnx-ptq Source: https://github.com/amd/Quark/tree/main/.claude/skills/quark-onnx-ptq Command: npx skills add https://github.com/amd/Quark --skill quark-onnx-ptq
This Skill removes the complexity of preparing, calibrating, and executing an end-to-end post-training quantization workflow for ONNX models with AMD Quark.
Ask the skill to run an end-to-end quantization workflow for your ONNX model and specify the target quantization format and any external weights file if present.
High-intent search queries and answers about installing and using this skill.
Post-training quantization for an ONNX model is executed end to end by generating calibration scripts, creating manifests, and running final optimized model generation steps for deterministic results.
AMD Quark supports common ONNX quantization formats including XINT8, A8W8, BFP16, and MXFP variants for vision models, as well as LLM weight-only compression.
Yes, you can use external weight files when quantizing ONNX models. The workflow requires support for ONNX inputs and optional external-weight files during the quantization process.
Calibration in an ONNX PTQ workflow operates through multi-step coordination that generates calibration scripts and manifests to plan quantization before executing the optimized model.
AMD Quark supports ONNX quantization for both vision models and LLMs, applying multi-step workflow coordination for vision scenarios and LLM weight-only compression tasks.
The best way to prepare an ONNX model for XINT8 or A8W8 quantization is through an end-to-end workflow that handles calibration planning, manifest generation, and execution confirmation.
Seeed-StudioConvert YOLO ONNX models to CVIMODEL for Sophgo CV181x TPU deployment.
zenobi-usDeploy scalable ML models with real-time inference, monitoring, and CI/CD pipelines.
amdGuide end-to-end post-training quantization of ONNX models with AMD Quark.
amdOrchestrate AMD Quark post-training quantization workflows for PyTorch and ONNX models.
