What is funsloth-train?

It is a specialized Claude Code skill designed to automate the creation of training notebooks and scripts for fine-tuning Large Language Models using the Unsloth library.

Which training techniques are supported?

The skill supports Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), Group Relative Policy Optimization (GRPO), ORPO, and KTO.

Can I use this for local training?

Yes, it generates scripts and notebooks compatible with local GPUs, as well as cloud providers like RunPod and Hugging Face Jobs.

Do I need deep ML knowledge to use this?

No, the 'Guide Me' mode walks you through every configuration step with explanations, while 'Sensible Defaults' provides production-ready settings automatically.

Which models can I fine-tune?

It supports most popular open-weight model families including Llama 3.1, Qwen, Gemma, Phi, Mistral, and DeepSeek.

FunSloth Unsloth Training Generator

Name: FunSloth Unsloth Training Generator
Author: chrisvoncsefalvay

bychrisvoncsefalvay

•

데이터 과학 및 ML

Generates optimized training notebooks and scripts for fine-tuning LLMs using the Unsloth framework.

소개

The funsloth-train skill streamlines the process of setting up Large Language Model (LLM) fine-tuning by generating tailored Unsloth notebooks and scripts. It simplifies complex configurations for SFT, DPO, and GRPO training through three intuitive interaction modes: production-ready defaults, guided setup for beginners, or interactive widget-based notebooks. By automating the selection of optimal parameters like LoRA rank, quantization, and learning rates across popular model families like Llama 3.1 and Mistral, it enables developers to move from raw data to a trained model faster while ensuring maximum VRAM efficiency.

주요 기능

Contextual guidance for deploying training to Hugging Face, RunPod, or local hardware.
Multi-mode configuration including 'Sensible Defaults' and a 'Guide Me' wizard.
Optimized 4-bit and 8-bit quantization setups for significant VRAM reduction.
4 GitHub stars
Automated generation of SFT, DPO, and GRPO training scripts and Jupyter notebooks.
Support for leading model families including Llama, Qwen, Mistral, and DeepSeek.

사용 사례

Optimizing training parameters for limited hardware resources like 8GB VRAM consumer GPUs.
Transitioning from standard SFT to preference optimization techniques like DPO or GRPO.
Setting up a production-ready Llama 3.1 fine-tuning environment with 4-bit quantization.

소개

주요 기능

Contextual guidance for deploying training to Hugging Face, RunPod, or local hardware.
Multi-mode configuration including 'Sensible Defaults' and a 'Guide Me' wizard.
Optimized 4-bit and 8-bit quantization setups for significant VRAM reduction.
4 GitHub stars
Automated generation of SFT, DPO, and GRPO training scripts and Jupyter notebooks.
Support for leading model families including Llama, Qwen, Mistral, and DeepSeek.

사용 사례

Optimizing training parameters for limited hardware resources like 8GB VRAM consumer GPUs.
Transitioning from standard SFT to preference optimization techniques like DPO or GRPO.
Setting up a production-ready Llama 3.1 fine-tuning environment with 4-bit quantization.