hqq-quantization

Half-Quadratic Quantization for LLMs without calibration data. Use when quantizing models to 4/3/2-bit precision without needing calibration datasets, for fast quantization workflows, or when deploying with vLLM or HuggingFace Transformers.

Quality

86%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

72%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The content is highly actionable with executable code and clean progressive disclosure into real one-level references, but it is held back by repetition of near-identical snippets and workflows that lack explicit validation checkpoints with error-recovery loops.

Suggestions

Collapse the repeated 4-bit BaseQuantizeConfig/HqqConfig and from_pretrained quantize-and-load examples into one canonical snippet, referencing it from later sections instead of restating it.

Add explicit validation checkpoints to the quantization workflows (e.g., a concrete quality acceptance criterion such as loss/perplexity delta or expected generation output, with a fix-and-retry step when it fails).

Move the bulk of the repeated HuggingFace integration variants (load pre-quantized, quantize-and-save, mixed precision) into references/advanced-usage.md, keeping only the primary path in SKILL.md.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient code without explaining basics Claude knows, but near-identical snippets (the 4-bit BaseQuantizeConfig/HqqConfig and from_pretrained quantize-and-load pattern) repeat 4-6 times, so not every token earns its place.	2 / 3
Actionability	Provides fully executable, copy-paste-ready code throughout (HQQLinear setup, HqqConfig with from_pretrained, vLLM quantization='hqq', PEFT LoraConfig, save/push_to_hub) plus specific pip commands, matching the executable-examples anchor.	3 / 3
Workflow Clarity	Workflows 1 and 2 are numbered and sequenced but verification is implicit (a 'Verify quality' print with no pass/fail criteria or fix-and-retry loop), so validation checkpoints are missing rather than explicit.	2 / 3
Progressive Disclosure	References to references/advanced-usage.md and references/troubleshooting.md are clearly signaled, both files exist, are one level deep, and the body appropriately keeps detail out of the overview.	3 / 3
	Total	10 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is strong: it concisely states a concrete capability, provides an explicit 'Use when' trigger with natural keywords, and occupies a clear niche unlikely to conflict with other skills. All four dimensions hit the top anchor.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions ('quantizing models to 4/3/2-bit precision without needing calibration datasets', 'deploying with vLLM or HuggingFace Transformers') in third-person voice, matching the multiple-specific-actions anchor.	3 / 3
Completeness	Explicitly answers both what ('Half-Quadratic Quantization for LLMs without calibration data') and when via an explicit 'Use when...' clause with concrete triggers, satisfying the both-what-and-when anchor.	3 / 3
Trigger Term Quality	Includes natural user phrases a person would say ('quantizing models', 'calibration datasets', 'fast quantization', 'vLLM', 'HuggingFace Transformers'), giving good coverage rather than just some keywords.	3 / 3
Distinctiveness Conflict Risk	A clear niche tied to calibration-free HQQ with named framework triggers makes it unlikely to fire for unrelated quantization skills, matching the distinct-niche anchor.	3 / 3
	Total	12 / 12 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: OpenLAIR/dr-claw
Path: skills/optimization/hqq/SKILL.md
Commit: 6da7f7c

Reviewed: about 6 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.