assemsabry

assemsabry / shade

Public

My website

assem.cloud
77
23
69% credibility
Found Feb 27, 2026 at 76 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Shade is an open-source platform that automatically removes censorship and safety filters from large language models using advanced optimization techniques while preserving their intelligence.

How It Works

1
🔍 Discover Shade

You stumble upon Shade online, a friendly tool promising to unlock AI assistants from annoying restrictions so they can help freely.

2
📥 Set it up

Download and install Shade on your computer with a simple command, like adding a new app.

3
🔑 Connect to models

Sign in to the model library so Shade can grab popular AI brains to work with.

4
✨ Free an AI

Pick your favorite AI model and let Shade automatically lift its censorship barriers while keeping its smarts intact.

5
💬 Chat away

Open the web chat or terminal to talk to your newly liberated AI, seeing side-by-side comparisons of before and after.

6
Share or save
💾
Save locally

Keep your uncensored AI on your computer for personal use anytime.

📱
Export to app

Send it to a chat app like Ollama for easy access on your phone or elsewhere.

🎉 AI unleashed!

Celebrate as your AI now answers openly without holding back, ready for honest conversations.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 76 to 77 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 shade?

Shade is a Python CLI tool that automatically strips censorship and safety filters from large language models using directional ablation—no fine-tuning or heavy training required. Point it at a Hugging Face model ID like `shade Qwen/Qwen2.5-1.5B-Instruct`, and it optimizes parameters with Optuna to maximize refusal removal while tracking KL divergence to preserve smarts. Developers get a web UI via `shade serve`, Ollama exports, benchmarking, and a diagnostic "doctor" for GPU tweaks.

Why is it gaining traction?

Unlike manual jailbreaks or finetuning scripts, Shade handles layer analysis, multi-GPU detection (CUDA, MPS, XPU), and self-healing deps out of the box, spitting out deployable uncensored models in minutes. The side-by-side original-vs-abliterated chat in the FastAPI web backend hooks tinkerers fast, plus one-click Ollama integration beats fiddly exports. At 76 stars, it's niche but buzzing in blox shade github circles for power shade github workflows.

Who should use this?

AI hobbyists running local LLMs who hate refusals on edgy prompts, researchers benchmarking uncensored behaviors, or indie devs prototyping shader-like flexible agents without cloud filters. Skip if you're shader minecraft 1.21 modders—it's for LLM liberation, not graphics.

Verdict

Grab it if you're deep in local model hacking; the automation shines despite beta quirks like sparse tests. 0.7% credibility score and low stars flag early maturity—fork and contribute to shade arena github stability. Solid 7/10 for serious uncensoring.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.