peft-fine-tuning

Parameter-efficient fine-tuning for LLMs using LoRA, QLoRA, and 25+ methods. Use when fine-tuning large models (7B-70B) with limited GPU memory, when you need to train <1% of parameters with minimal accuracy loss, or for multi-adapter serving. HuggingFace's official library integrated with transformers ecosystem.

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 body is a highly actionable cookbook of executable PEFT examples with clean progressive disclosure into two reference files. It loses points only on conciseness (volume) and workflow clarity (missing explicit validation checkpoints in the training flows).

Suggestions

Move the performance benchmark tables and the secondary method examples (IA3, Prefix Tuning) into references/advanced-usage.md to slim the main body.

Add explicit validation checkpoints to the training workflow (e.g., verify loss is decreasing and adapter is active before saving) to form a clear feedback loop.

Consolidate the 'Common issues' and 'Best practices' sections, since some points duplicate guidance already shown in the inline code comments.

Dimension	Reasoning	Score
Conciseness	Content is mostly lean executable code with useful inline comments and no padding about concepts Claude already knows, but the sheer volume (multiple full integration examples, performance benchmark tables, and a methods comparison) could be tightened or moved to references.	2 / 3
Actionability	Provides fully executable, copy-paste-ready code for LoRA, QLoRA, IA3, prefix tuning, loading/merging, multi-adapter serving, and integrations (TRL, Axolotl, vLLM), with specific config values throughout.	3 / 3
Workflow Clarity	The quick-start sequences (load model → configure → apply → train → save) are clear and include a print_trainable_parameters() sanity check, but the training workflows lack explicit validation checkpoints or feedback loops between steps.	2 / 3
Progressive Disclosure	SKILL.md is a well-organized overview that signals one-level-deep references to the real files references/advanced-usage.md (DoRA, LoftQ, custom modules) and references/troubleshooting.md, neither of which nest further.	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 specific, uses natural trigger terms, and explicitly covers both what the skill does and when to use it. Third-person voice is maintained throughout with no over-claims.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete capabilities and methods — 'LoRA, QLoRA, and 25+ methods', 'train <1% of parameters', 'multi-adapter serving' — rather than vague abstractions, matching the top anchor.	3 / 3
Completeness	Explicitly answers both what ('Parameter-efficient fine-tuning ... using LoRA, QLoRA, and 25+ methods') and when ('Use when fine-tuning large models (7B-70B) with limited GPU memory ... or for multi-adapter serving').	3 / 3
Trigger Term Quality	Uses natural terms a user would actually say when needing this skill — 'fine-tuning', 'LoRA', 'QLoRA', 'limited GPU memory', 'multi-adapter serving' — giving good coverage of common phrasings.	3 / 3
Distinctiveness Conflict Risk	Occupies a clear niche (PEFT/fine-tuning) with distinctive named methods and triggers unlikely to fire for unrelated skills.	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: OpenRaiser/NanoResearch
Path: skills/vendor-ai-research/peft/SKILL.md
Commit: 7144364

Reviewed: about 13 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.