Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Expert guidance for fast fine-tuning with Unsloth - 2-5x faster training, 50-80% less memory, LoRA/QLoRA optimization
.claude/skills/graniet-unsloth/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -35% | 0% |
This skill is repo-local and stays inactive until explicitly activated.
When the original instructions refer to legacy tool names, use these Kheish mappings:
terminal => bashweb_extract => web_fetch, plus web_search when discovery is neededsearch_files => grep_search and glob_searchbrowser_* tools require a browser-capable surfaced tool or MCP; if none is available, use the closest available surface and say so explicitlyWhen the instructions mention local helper files, resolve them from ${KHEISH_SKILL_DIR}.
Comprehensive assistance with unsloth development, generated from official documentation.
This skill should be triggered when:
Quick reference patterns will be added as you use the skill.
This skill includes comprehensive documentation in references/:
Use view to read specific reference files when detailed information is needed.
Start with the getting_started or tutorials reference files for foundational concepts.
Use the appropriate category reference file (api, guides, etc.) for detailed information.
The quick reference section above contains common patterns extracted from the official docs.
Organized documentation extracted from official sources. These files contain:
Add helper scripts here for common automation tasks.
Add templates, boilerplate, or example projects here.
To refresh this skill with updated documentation:
<!-- Trigger re-upload 1763621536 -->
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 2,062 | 3,687 | +79% | 1 | 1 | 0% | 185 | 772 | +317% | 0 | 0 | — |
case-02 | fail→pass | 9,653 | 2,103 | -78% | 1 | 1 | 0% | 1,654 | 863 | -48% | 0 | 0 | — |
case-03 | pass→pass | 2,908 | 1,934 | -33% | 1 | 1 | 0% | 473 | 876 | +85% | 0 | 0 | — |
case-04 | fail→pass | 6,446 | 2,308 | -64% | 1 | 1 | 0% | 1,112 | 851 | -23% | 0 | 0 | — |
case-05 | fail→pass | 4,430 | 2,269 | -49% | 1 | 1 | 0% | 703 | 831 | +18% | 0 | 0 | — |
case-06 | pass→pass | 9,623 | 2,849 | -70% | 1 | 1 | 0% | 1,528 | 1,020 | -33% | 0 | 0 | — |
case-07 | fail→pass | 9,493 | 1,499 | -84% | 1 | 1 | 0% | 1,483 | 757 | -49% | 0 | 0 | — |
case-08 | fail→pass | 7,747 | 1,902 | -75% | 1 | 1 | 0% | 1,267 | 821 | -35% | 0 | 0 | — |
case-09 | fail→pass | 6,592 | 3,884 | -41% | 1 | 1 | 0% | 1,096 | 1,040 | -5% | 0 | 0 | — |
case-10 | pass→pass | 7,857 | 1,465 | -81% | 1 | 1 | 0% | 1,299 | 703 | -46% | 0 | 0 | — |
case-11 | fail→pass | 11,761 | 2,102 | -82% | 1 | 1 | 0% | 2,118 | 880 | -58% | 0 | 0 | — |
case-12 | pass→pass | 3,117 | 4,549 | +46% | 1 | 1 | 0% | 553 | 1,370 | +148% | 0 | 0 | — |
case-13 | pass→pass | 2,918 | 5,524 | +89% | 1 | 1 | 0% | 511 | 1,201 | +135% | 0 | 0 | — |
case-19 | pass→fail | 14,113 | 4,451 | -68% | 1 | 1 | 0% | 2,809 | 765 | -73% | 0 | 0 | — |
case-14 | pass→pass | 11,830 | 4,621 | -61% | 1 | 1 | 0% | 2,099 | 1,288 | -39% | 0 | 0 | — |
case-15 | pass→fail | 8,761 | 4,105 | -53% | 1 | 1 | 0% | 1,830 | 789 | -57% | 0 | 0 | — |
case-16 | pass→pass | 2,278 | 2,545 | +12% | 1 | 1 | 0% | 380 | 984 | +159% | 0 | 0 | — |
case-17 | pass→fail | 6,945 | 4,800 | -31% | 1 | 1 | 0% | 1,266 | 740 | -42% | 0 | 0 | — |
case-18 | pass→fail | 7,761 | 3,864 | -50% | 1 | 1 | 0% | 1,415 | 709 | -50% | 0 | 0 | — |
case-20 | pass→pass | 3,123 | 3,115 | -0% | 1 | 1 | 0% | 588 | 1,103 | +88% | 0 | 0 | — |
case-21 | fail→pass | 6,695 | 4,472 | -33% | 1 | 1 | 0% | 1,285 | 1,309 | +2% | 0 | 0 | — |
case-22 | pass→pass | 3,096 | 2,901 | -6% | 1 | 1 | 0% | 551 | 928 | +68% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 18 counted toward the lift figure. The other 4 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +18 percentage points is the difference between those two pass rates over the 18 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.