unsloth-quantization
Unsloth utilizes advanced quantization techniques to reduce the memory footprint of LLM fine-tuning. This includes "Dynamic 4-bit" loading (protecting sensitive layers), FP8 training for modern GPUs, and the use of 8-bit optimizers to save gigabytes of VRAM.
npx skills add majiayu000/claude-skill-registry --skill unsloth-quantization-cuba6112-skillfactory-6a9fb307 --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
--- name: unsloth-quantization description: Utilizing Dynamic 4-bit quantization, FP8 training, and 8-bit optimizers to minimize VRAM usage without sacrificing accuracy. Triggers: quantization, dynamic 4-bit, fp8, bitsandbytes, adamw_8bit, qat. --- ## Overview Unsloth utilizes advanced quantization techniques to reduce the memory footprint of LLM fine-tuning. This includes "Dynamic 4-bit" loading (protecting sensitive layers), FP8 training for modern GPUs, and the use of 8-bit optimizers to save gigabytes of VRAM. ## When to Use - When training on GPUs with limited VRAM (e.g., 8GB, 12GB, or 16GB). - When aiming for the fastest possible training speeds on H100 or RTX 40 series GPUs. - When trying to balance model size and reasoning performance. ## Decision Tree 1. Is your GPU RTX 40 series or newer (Ada/Hopper)? - Yes: Use FP8 Dynamic for 2x faster training. - No: Use BF16/FP16. 2. Running out of VRAM? - Yes: Ensure `load_in_4bit=True` and use `adamw_8bit` optimizer. 3. Is accuracy dropping significantly? - Yes: Use "Dynamic" variants that protect the first and last layers. ## Workflows ### FP8 Training Configuration 1. Select a model variant ending in '-FP8-Dynamic'. 2. Configure t
- Overview
- When to Use
- Decision Tree
- Workflows
- FP8 Training Configuration
- VRAM-Constrained Training Setup
- Non-Obvious Insights
- Evidence
- Scripts
- Dependencies
- References
What does the unsloth-quantization skill do?
Unsloth utilizes advanced quantization techniques to reduce the memory footprint of LLM fine-tuning. This includes "Dynamic 4-bit" loading (protecting sensitive layers), FP8 training for modern GPUs, and the use of 8-bit optimizers to save gigabytes of VRAM.
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill unsloth-quantization-cuba6112-skillfactory-6a9fb307 --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.