openvinotoolkit - Read src/frontends/gguf/docs/debugging_accuracy.md — the coarse-to-fine bisection strategy, the ggml-CPU oracle technique, the catalogue of bug archetypes with their generalizable lessons, the debug env vars, and the checklist.
- Apply its one governing rule before anything else: every accuracy claim is a comparison against the real llama.cpp CPU implementation, never a hand-derived reference. If you are about to write out an op's math to form an expectation, generate it from ggml instead.
- Work the bisection steps in order and do not open a debugger until they have cornered the bug. Check the bug archetypes first when a cosine cliff or dynamic-shape crash points near one.
- After any fix to a shared path — especially a VIEW or
op_casepredicate — re-run the other supported architectures and confirm their classification and output are unchanged.
Ov Gguf Debug Accuracy
Debug wrong or degraded output from a GGUF model running through the OpenVINO GGUF frontend or the ggml-openvino backend. Use when a GGUF model converts but produces garbage, repeated or drifting tokens, a cosine-similarity cliff between layers, output that diverges from llama.cpp CPU, or a broadcast/shape crash that appears only at decode or only for one architecture. Do NOT use for ops that fail to convert at all ("Translation for operation type ... is not implemented"), or for build and CMake failures.
Ov Gguf Debug Accuracy by openvinotoolkit · 503b981
npx skillmds@latest add openvinotoolkit/ov-gguf-debug-accuracy File contents
---name: ov-gguf-debug-accuracydescription: Debug wrong or degraded output from a GGUF model running through the OpenVINO GGUF frontend or the ggml-openvino backend. Use when a GGUF model converts but produces garbage, repeated or drifting tokens, a cosine-similarity cliff between layers, output that diverges from llama.cpp CPU, or a broadcast/shape crash that appears only at decode or only for one architecture. Do NOT use for ops that fail to convert at all ("Translation for operation type ... is not implemented"), or for build and CMake failures.---1. Read [src/frontends/gguf/docs/debugging_accuracy.md](../../../src/frontends/gguf/docs/debugging_accuracy.md) — the coarse-to-fine bisection strategy, the ggml-CPU oracle technique, the catalogue of bug archetypes with their generalizable lessons, the debug env vars, and the checklist.2. Apply its one governing rule before anything else: every accuracy claim is a comparison against the **real llama.cpp CPU implementation**, never a hand-derived reference. If you are about to write out an op's math to form an expectation, generate it from ggml instead.3. Work the bisection steps in order and do not open a debugger until they have cornered the bug. Check the bug archetypes first when a cosine cliff or dynamic-shape crash points near one.4. After any fix to a shared path — especially a VIEW or `op_case` predicate — re-run the other supported architectures and confirm their classification and output are unchanged.
openvinotoolkit/openvino/tree/main/.claude/skills/ov-gguf-debug-accuracy commit 503b9815e0
Frequently asked questions
Run npx skillmds@latest add openvinotoolkit/ov-gguf-debug-accuracy in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.
Debug wrong or degraded output from a GGUF model running through the OpenVINO GGUF frontend or the ggml-openvino backend. Use when a GGUF model converts but produces garbage, repeated or drifting tokens, a cosine-similarity cliff between layers, output that diverges from llama.cpp CPU, or a broadcast/shape crash that appears only at decode or only for one architecture. Do NOT use for ops that fail to convert at all ("Translation for operation type ... is not implemented"), or for build and CMake failures. It is listed under Web & Frontend on SkillMD.
This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.
This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.
Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.
openvinotoolkit (@openvinotoolkit) published this skill. Their other Agent Skills are listed on their SkillMD profile.