anaksor-ocr
Sokheng · Kit
Reads Khmer and English text from documents, receipts and photos of signs, using a detector for text lines and a recogniser for each line. Suits developers who…
New
Overview
Modifies local language models by adjusting internal weights to remove refusals, reduce slop, or change stylistic behavior. Targeted at developers and hobbyists working with custom LLMs.
Heretic is an open-source project on GitHub in the Developer category, listed by Mindket. It is free. This is version 1, last updated on . Install it from GitHub as its README says. Get Mindket’s setup notes on your Mindket account, as a zip.
At a glance
- You need
- Python 3.10 or newer
- PyTorch 2.6 or newer
- Compatible hardware
- Start with
pip install -U heretic-llm heretic Qwen/Qwen3-4B-Instruct-2507
About
What it is
Heretic modifies language models by adjusting internal weights to steer behavior. It performs these changes without training, and it supports tasks like removing refusals, reducing slop, or altering emotional alignment.
Who it suits
It is for developers and hobbyists who want to modify models on their own hardware.
Where to use it
You use it on your local computer to process models you run locally.
How to set it up
- Python 3.10 or newer
- PyTorch 2.6 or newer
- Compatible hardware
- Prepare your Python environment.
- Install the package via your terminal.
pip install -U heretic-llm heretic Qwen/Qwen3-4B-Instruct-2507
How to use it
You run commands in your terminal to process a model. By default, it removes censorship and suppresses refusals. You can specify a configuration file to change its goals.
For newcomers and experts
You do not need to understand how models work internally to use this. You can remove censorship from your first model in seconds. If you are new, try running it on a dense model to observe how it changes the output.
Good to know
- Supports dense, multimodal, MoE, and hybrid models.
- Performs on-the-fly 4-bit quantization with bitsandbytes.
- Includes reproducibility features to recreate models using a manifest.
- Licensed under the GNU Affero General Public License.
README and files
Heretic
Modifies local language models by adjusting internal weights to remove refusals, reduce slop, or change stylistic behavior. Targeted at developers and hobbyists working with custom LLMs.
Made by p-e-w on GitHub. These notes are Minda's, written from its README; the project's own files and license are in its repository.
About
What it is: Heretic modifies language models by adjusting internal weights to steer behavior. It performs these changes without training, and it supports tasks like removing refusals, reducing slop, or altering emotional alignment.
Who it suits: It is for developers and hobbyists who want to modify models on their own hardware.
Where to use it: You use it on your local computer to process models you run locally.
How to set it up:
- Python 3.10 or newer
- PyTorch 2.6 or newer
- Compatible hardware
- Prepare your Python environment.
- Install the package via your terminal.
pip install -U heretic-llm
heretic Qwen/Qwen3-4B-Instruct-2507How to use it: You run commands in your terminal to process a model. By default, it removes censorship and suppresses refusals. You can specify a configuration file to change its goals.
For newcomers and experts: You do not need to understand how models work internally to use this. You can remove censorship from your first model in seconds. If you are new, try running it on a dense model to observe how it changes the output.
Good to know:
- Supports dense, multimodal, MoE, and hybrid models.
- Performs on-the-fly 4-bit quantization with bitsandbytes.
- Includes reproducibility features to recreate models using a manifest.
- Licensed under the GNU Affero General Public License.
On GitHub, 4 November 2026
- Stars: 38,841 (701 that day)
- Forks: 4,387
- Watching: 199
- Open issues and pull requests: 62
- Language: Python
- License: AGPL-3.0
- Topics: abliteration, anti-censorship, anticensorship, censorship, censorship-resistance, llm, transformer
- Trending: #2 repository of the day on GitHub trending (4 November 2026)
Links
- Code: https://github.com/p-e-w/heretic
- Website: https://heretic-project.org/
- Releases: https://github.com/p-e-w/heretic/releases
- Issues: https://github.com/p-e-w/heretic/issues
- Star history: https://www.star-history.com/#p-e-w/heretic&Date
Get the code
git clone https://github.com/p-e-w/heretic.gitThen follow the setup in its README. Nothing in this kit runs by itself.
README.md 2 KB
What’s new
Reviews
What people say
Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search and on Trendshift. Mindket doesn’t write or check them. Open one to read it where it was posted.
razsterHeretic 2.0 Released: Training-free language model modificationhttps://github.com/p-e-w/ -p-e-w- Posted this from Reddit. "Dear fellow Llamas, after almost 5 months of development, I finally have the pleasure of announcing to you the immediate availability of the biggest Heretic release ever, Heretic 2.0: https://github.com/p-e-w/heretic Heretic is the…22 points · 4 replies@ai-trend-to-know.bsky.socialp-e-w/heretic 💻 Python #Python #OpenSource #AIAgents #GitHub 🔗 trendtoknow.com/p-e-w/heretic4 likes@venv_kHeretic 🧠 An open-source tool for removing LLM censorship without additional training. It uses abliteration to modify model weights and Optuna to optimize the process. https://t.co/5u8vFDOjWy https://t.co/YLc1qsOZFc #AI #LLM #OpenSourcefreakynitAstra can read and write weights to improve QwenThis was the first thing that came to my mind as well. For the unknowns, this is a pretty/popular project for the same task: https://github.com/p-e-w/heretic You can search huggingface for models with word "heretic" in them... https://huggingface.co/models?sort=created&search=hereticHenrique OlifiersI strongly suggest we start preparing for the worst. github.com/p-e-w/heretic3 likes · 1 reply@JubayarAdnan38,000+ devs starred this free tool 😳 Heretic removes censorship from AI models you run at home. - No training needed - Works on a gaming PC - Can also cut "slop" and fake praise - 1000s of models already made with it Link in the first comment 👇 https://t.co/uESFoVdSNewalrus01Memdoor: A team of 5 Qwen3.8-27B models beats Fable 5 on LiveCodeBench HardYes, it'll likely just result in some people standing up a more dedicated purpose community hosting site of abliterated/uncensored open weight GGUF files. In a theoretical scenario where HF bans abliterated/uncensored models I could very well see the same people who are producing those now simply…Mark IgraAI "alignment" will not prevent bad actors from using open models for bad acts: "Heretic is a tool that removes censorship (aka "safety alignment") from transformer-based language models...anyone who knows how to run a command-line program can use Heretic" github.com/p-e-w/heretic4 likes@VT_zero1Top AI repos on GitHub today | Wed, Nov 4 • vercel-labs/agent-desktop: A command line tool that lets an AI agent use your computer. It can click around and read what is on screen. https://t.co/vNU6WOq8b3 • Donchitos/Claude-Code-Game-Studios: Turns Claude Code into a small game studio, with 49…Retro_DevClaude Opus 5.5> Distillation attacks, in which attackers use thousands of fake accounts to extract a model’s capabilities at industrial scale, create safety and national security risks. Distillation allows bad actors to create highly capable models without the safeguards we build into Claude. Our September 2026…KeithThere is a fully automated tool that does it. People have weights with the refusal stripped out published within hours of any new model drop. Even w/ stock weights, you can modify your model running code to suppress that activation vector. Barn door fully off the hinges. github.com/p-e-w/heretic3 likes@JkwuRmsl试了下 heretic 不训练直接改本地模型行为 去审查 压 AI 味 治谄媚 GitHub 三万八千星 你更信这种拧权重 还是重训一版 https://t.co/5SFs7aubBw https://t.co/rm4ce3ESHn
More from Mindket
See all from MindketQwen2.5-7B-Instruct-GGUF by Qwen
Qwen · Kit
Qwen2.5-7B-Instruct as GGUF files in eight quantization sizes, to chat with locally through llama.cpp. For people who want a capable 7.6B model on their own co…
New
ACE-Step 1.5 XL SFT
ACE-Step · Kit
Generates music from text with a 4B diffusion decoder, 50 steps and guidance control, aiming for higher audio quality than the base. For creators with a GPU of…
New
Similar in Developer
See all in DeveloperClaude Code Skills & Plugins
alirezarezvani · Skill
Adds 411 ready-made skills for Claude Code and other coding agents, grouped by role from engineering and product to marketing, finance and compliance. Installs…
New29.8k
buzz
block · Kit
Self-hostable workspace where people and AI agents share the same rooms, built on a Nostr relay that logs every action as a signed event. For teams that want t…
New37.3k
plannotator
backnotprop · Kit
Opens your AI coding agent's plans, specs and code changes in a browser so you can annotate or comment on them and send the feedback straight back to the agent.
New10.5k