AI & LLM Engineering · amd/Quark

Run, resume, monitor, diagnose, and report Quark Quant-Perf workflows for PyTorch and HuggingFace transformers models.

amd/Quark

Author a new ShapeShifter graph-transformation pass for AMD Quark (ONNX or PyTorch) so it conforms to the pass framework's conventions and auto-registers.

amd/Quark

Collect and normalize environment facts (OS, Python, GPU, CUDA/ROCm, container state) before Quark installation or PTQ planning.

amd/Quark

Install or verify the AMD Quark package and its dependencies.

amd/Quark

L3 recipe that runs quark.onnx.AutoSearchPro end-to-end on a user .onnx model: intake → preset selection (or custom search space) → calibration / eval data reader → standalone autosearch script…

amd/Quark

Diagnose failed Quark ONNX installation, calibration, quantization, custom-op compilation, or export attempts.

amd/Quark

Install or verify the correct ONNX Runtime build (and the onnx package) for a user's accelerator backend before Quark's ONNX-to-ONNX flow.

amd/Quark

Inspect a target ONNX model and prepare metadata for Quark ONNX PTQ planning.

amd/Quark

End-to-end ONNX PTQ workflow for AMD Quark — from a .onnx file (and calibration data) to a quantized .onnx output.

amd/Quark

Build a Quark ONNX PTQ quantization plan from modelanalysis.json and user intent.

amd/Quark

Validate Quark ONNX quantization output using four lightweight checks: auxiliary file copy alignment, expected non-quantized initializer MD5 byte-identity (inline rawdata + external-data byte…

amd/Quark

Route Quark ONNX user goals to the correct atomic skill. An agent skill from amd/Quark.

amd/Quark

Apply existing ShapeShifter graph passes to an .onnx model via the quark-cli shapeshifter CLI or a ShapeShifter YAML.

amd/Quark

Detect upstream Quark ONNX changes that affect the ONNX skill family and classify required updates.

amd/Quark

Partition an ONNX model graph into named functional subgraphs and emit a subgraphpartition.json file.

amd/Quark

Diagnose failed Quark installation, PTQ execution, script generation, or export attempts.

amd/Quark

Compare skill contracts and guidance against current Quark documentation and source entry points.

amd/Quark

Prepare export and downstream evaluation handoff for a planned or completed Quark PTQ run.

amd/Quark

Low-memory file2file quantization for very large safetensors LLMs that cannot be loaded whole.

amd/Quark

Install or verify the correct PyTorch build for a user's accelerator backend before Quark installation.

amd/Quark

L3 recipe that runs a Torch LLM PTQ end-to-end for AMD Quark — for PyTorch / HuggingFace transformers models (safetensors input): quantize → validate → evaluate.

amd/Quark

Torch LLM PTQ workflow for AMD Quark — from model selection to quantized output.

amd/Quark

Inspect a target model and prepare metadata for Quark PTQ planning.

amd/Quark

Build a Quark Torch LLM PTQ quantization plan from model analysis and user intent.

amd/Quark

Validate Quark quantization output using four lightweight checks: auxiliary file copy alignment, excluded tensor MD5 byte-identity, config.json deep comparison after stripping quantization keys, and…

amd/Quark

Route Quark user goals to the correct atomic skill or workflow.

amd/Quark

Route Quark user goals to the correct atomic skill or workflow.

amd/Quark

Detect upstream Quark changes that affect the skill system and classify required updates.

amd/Quark

Validate workspace shape, model paths, output directories, and repo structure before downstream Quark skills proceed.

amd/Quark

Installs or verifies AMD Quark and ensures the selected Python environment has an accelerator-matched PyTorch.

amd/Quark

Runs an end-to-end AMD Quark post-training quantization workflow for PyTorch / Hugging Face LLMs: inspect a Hub or local model, choose a quantization plan, create reproducible artifacts, request…

amd/Quark

End-to-end LLM accuracy evaluation on AMD ROCm (ROCm-only) — container setup, vLLM/SGLang/ATOM serving, lm-eval / lighteval / evalscope benchmarks.

amd/Quark

Shrink a HuggingFace safetensors model to 1 hidden layer for fast debugging without loading the full model into memory.

amd/Quark