unsloth

Expert guidance for fast fine-tuning with Unsloth - 2-5x faster training, 50-80% less memory, LoRA/QLoRA optimization

Quality

36%

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

Fix and improve this skill with Tessl

tessl review fix ./bundled/skills/unsloth/SKILL.md

Quality

Content

15%

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

The body is an unpopulated auto-generated template: it contains no executable guidance, no real workflow, and references several reference files that do not exist, making it largely non-functional as actionable skill content.

Suggestions

Add at least one concrete, copy-paste fine-tuning example (e.g. a minimal Unsloth SFT snippet with model loading, LoRA config, and Trainer setup) instead of the empty 'Quick Reference'.

Document a real multi-step workflow (prepare dataset -> load model + LoRA -> train -> save merged adapter) with a validation/checkpoint step, rather than only generic 'When to Use' bullets.

Fix the broken references: remove or replace mentions of 'getting_started', 'tutorials', 'api', and 'guides' with the files that actually exist (llms-txt.md / llms.md), or generate the missing files.

Dimension	Reasoning	Score
Conciseness	The body is short, but it is padded with generic auto-generated boilerplate ('Comprehensive assistance... generated from official documentation', 'This skill was automatically generated...', the restating 'Notes' section) that adds no value Claude doesn't already infer; it could be tightened to drop the template filler.	2 / 3
Actionability	There is no concrete code, command, or example anywhere — the Quick Reference explicitly says 'patterns will be added as you use the skill' (empty), and the only instruction is the vague 'Use `view` to read specific reference files'; it describes rather than instructs.	1 / 3
Workflow Clarity	Fine-tuning is inherently a multi-step process (data prep, model config, training, eval), yet the body provides no sequenced steps, no validation checkpoints, and no feedback loops; it only offers generic 'When to Use' bullets and pointers to nonexistent files.	1 / 3
Progressive Disclosure	The body references 'getting_started', 'tutorials', 'api', and 'guides' reference files that do not exist in references/ (which contains only index.md, llms-txt.md, llms.md, and an empty llms-full.md), so the navigation is broken; only llms-txt.md is a real, correctly signaled reference.	1 / 3
	Total	5 / 12 Passed

Description

57%

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 conveys a clear niche and some concrete method keywords, but it is padded with benefit claims ('Expert guidance', '2-5x faster', '50-80% less memory') and lacks any explicit 'Use when' trigger clause, leaving the activation context implied rather than stated.

Suggestions

Replace 'Expert guidance for' with concrete verbs, e.g. 'Fine-tune LLMs with Unsloth using LoRA/QLoRA for faster, memory-efficient training.'

Add an explicit trigger clause: 'Use when the user wants to fine-tune, train with LoRA/QLoRA, or reduce training memory/time for models like Llama, Mistral, Gemma, or Qwen.'

Drop unverifiable quantified benefit claims ('2-5x faster', '50-80% less memory') unless tied to specific, dated benchmarks.

Dimension	Reasoning	Score
Specificity	Names the domain ('fast fine-tuning with Unsloth') and methods ('LoRA/QLoRA optimization'), but leads with the fluffy phrase 'Expert guidance' and lists performance benefits (2-5x faster, 50-80% less memory) rather than concrete user-facing actions like 'train', 'fine-tune', 'optimize', or 'quantize'.	2 / 3
Completeness	It answers 'what' (fast fine-tuning with Unsloth, LoRA/QLoRA optimization) but the 'when' is entirely absent — there is no explicit trigger guidance, which per the judging guidelines caps completeness at 2.	2 / 3
Trigger Term Quality	Includes relevant natural terms ('fine-tuning', 'Unsloth', 'LoRA', 'QLoRA') that ML practitioners would say, but lacks a 'Use when...' clause and is missing common variations like 'train', 'peft', 'adapter', or 'quantize' that would broaden coverage.	2 / 3
Distinctiveness Conflict Risk	'Fine-tuning with Unsloth' plus 'LoRA/QLoRA optimization' defines a clear, narrow niche tied to a specific library, making it unlikely to be selected for unrelated skills.	3 / 3
	Total	9 / 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: foryourhealth111-pixel/Vibe-Skills
Path: bundled/skills/unsloth/SKILL.md
Commit: f627ab5

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.