Instructions to use llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Use Docker

docker model run hf.co/llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Ollama
How to use llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF with Ollama:
```
ollama run hf.co/llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M
```
Unsloth Desktop

How to use llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF with Docker Model Runner:
```
docker model run hf.co/llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M
```

Lemonade

How to use llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "llmfan46/Qwen3.5-35B-A3B-ultra-uncensored-heretic-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Are you planning a heretic version of Qwen3.5‑122B?

by Swat007 - opened Mar 3

Discussion

Swat007

Mar 3

Thanks for releasing the uncensored versions. Do you plan to make one for Qwen3.5‑122B (GGUF) as well?

llmfan46

Owner Mar 3

•

edited Mar 3

Thanks for releasing the uncensored versions. Do you plan to make one for Qwen3.5‑122B (GGUF) as well?

Yes and no, I actually tried twice to do that for Qwen3.5‑122B, the issue is the enormous size, this model on it's own is about 250GB, so you would need more than 250GB of VRAM to do it, right now the biggest VRAM possible is on the H200 which has 141GB of VRAM and cost over 30000-35000 USD in the US and about 35000 EUR in the EU and you would need 2 of these (141 GB x 2 = 282 GB of VRAM), So the answer is yes I would love to do it for Qwen3.5‑122B and Qwen3.5-397B and other big models too, I just need to figure how I am going to do it and there are a few other ways that I could try to do it, but I would need to open up donations with Ko-Fi or Patreon or something to help cover the costs.

SassyDiffusion

Mar 15

Can it be done on a 512GB Mac?

llmfan46

Owner Mar 15

•

edited Mar 15

Can it be done on a 512GB Mac?

Well I don't know much about Mac support, but I am assuming you are talking about the Mac Studio with the M3 Ultra chip and 512GB of Unified memory? This has enough RAM, the issue is the support, Mac doesn't have CUDA support (Mac has "Metal") and there are currently as far as I understand no Mac support for transformers and well you need transformers, what could maybe work (not sure) is if you had a computer that has CUDA and transformers and you connected that PC to the the Mac and used the Mac only as a compute workhorse, that could work if such a setup was possible, the issue is that I have no idea how fast/slow this would be nor if it would even work.

SassyDiffusion

Mar 16

I have a 128GB M4 Max and transformers does work for me, what didn't work was accelerate and bitsandbytes. Not sure if these are required (usually I can make things work without them)?

llmfan46

Owner Mar 16

I have a 128GB M4 Max and transformers does work for me, what didn't work was accelerate and bitsandbytes. Not sure if these are required (usually I can make things work without them)?

Yes they are required, you need:

accelerate: 1.13 https://pypi.org/project/accelerate/1.13.0/

and

bitsandbytes: 0.49 https://pypi.org/project/bitsandbytes/0.49.2/

SassyDiffusion

Mar 16

•

edited Mar 16

I think accelerate and bitsandbytes are just for saving memory (quantization and gradient accumulation etc), I've ran multiple repos without them on macOS which normally use them on nvidia cards because nvidia cards have little memory compared to macs. For bitsandbytes I usually just remove it from dependencies and for accelerate there's usually a way to run without it, since it's just like a wrapper that provides optimization. Maybe worth a shot.

Edit: Accelerate is supported on mac, but people usually use it with bitsandbytes to run models quantized, which won't work on mac, so either run with accelerate and disable quantization or just run without it. If you're not going to do this quantized anyway then I don't think there's any loss by running without these.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment