TYH-labs

TYH-labs / unsloth-buddy

Public

Zero-friction LLM fine-tuning skill for Claude Code. Unsloth on NVIDIA/CUDA · mlx-tune on Apple Silicon. Automates env setup, LoRA training (SFT, DPO, GRPO, vision), evaluation, and export end-to-end. Part of the Gaslamp AI development platform.

gaslamp.devunsloth apple-silicon claude-code dpo fine-tuning gaslamp
17
1
89% credibility
Found Mar 20, 2026 at 17 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
HTML
AI Summary

unsloth-buddy is a conversational agent that automates fine-tuning AI language models on user data through a guided interview, data preparation, training, evaluation, and export process across various hardware.

How It Works

1
💡 Find your AI training buddy

You stumble upon unsloth-buddy while searching for an easy way to teach an AI your own questions and answers, like customer support chats.

2
🔌 Add it to your AI helper

With a quick command in your AI chat tool, you plug it in like adding a new app to your phone.

3
🗣️ Describe your dream AI

You chat naturally: 'Make a helper for my FAQ file on my laptop,' and it asks friendly questions to nail down exactly what you need.

4
📁 It sorts your information

It grabs your files, reshapes them into perfect lessons, and confirms everything looks good.

5
🚀 Watch it learn live

It uses your computer's power or free cloud help to train, opening a colorful screen showing progress bars and graphs updating in real time.

6
👀 Compare before and after

It runs tests side-by-side: the old generic AI vs. your smart new one, proving how much better it handles your world.

🎉 Launch your personal AI

You get a ready-to-go custom brain packed neatly, easy to run on your phone, computer, or anywhere, knowing your stuff inside out.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 17 to 17 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is unsloth-buddy?

unsloth-buddy is a Claude Code skill that automates end-to-end LLM fine-tuning, from env setup to export. Tell it your goal—like fine-tuning a summarizer on customer support CSVs—and it handles data formatting, picks SFT/DPO/GRPO/vision methods, trains via Unsloth on NVIDIA or mlx-tune on Apple Silicon, runs evaluation, and exports deployable models. Built as HTML with multilingual docs, it's part of the Gaslamp development platform for agentic workflows.

Why is it gaining traction?

It skips manual boilerplate: auto-detects hardware, reformats messy data, shows base vs. fine-tuned outputs, and spins up a live dashboard at localhost:8080. Developers love the conversational flow—one chat yields a GGUF for Ollama—plus Colab offload for free GPUs and seamless integration with Claude Code or Gemini CLI. No more env mismatches or "which LoRA rank?" debates.

Who should use this?

ML engineers prototyping preference tuning (DPO/GRPO) on local hardware; indie devs on Apple Silicon Macs fine-tuning vision models without cloud costs; teams in Gaslamp or Claude Code workflows needing quick SFT on FAQs/code datasets. Skip if you prefer full control over trainers.

Verdict

Grab it for frictionless fine-tuning experiments—0.9% credibility score and 17 stars signal early maturity, but solid docs and MIT license make it low-risk to try. Pairs best with agentic dev tools; monitor for broader adoption.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.