Instructions to use deucebucket/Qwen3.6-35B-A3B-Cerebellum-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 deucebucket/Qwen3.6-35B-A3B-Cerebellum-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 deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M
# Run inference directly in the terminal:
llama cli -hf deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M
# Run inference directly in the terminal:
llama cli -hf deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_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 deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M
# Run inference directly in the terminal:
./llama-cli -hf deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_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 deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M

Use Docker

docker model run hf.co/deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M

LM Studio
Jan

vLLM

How to use deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "deucebucket/Qwen3.6-35B-A3B-Cerebellum-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": "deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M

Ollama
How to use deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF with Ollama:
```
ollama run hf.co/deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M
```
Unsloth Desktop

How to use deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_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": "deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF with Docker Model Runner:
```
docker model run hf.co/deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M
```

Lemonade

How to use deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M

Run and chat with the model

lemonade run user.Qwen3.6-35B-A3B-Cerebellum-GGUF-Q3_K_M

List all available models

lemonade list

Hermes Agent

How to use deucebucket/Qwen3.6-35B-A3B-Cerebellum-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 deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_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 deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_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 "deucebucket/Qwen3.6-35B-A3B-Cerebellum-GGUF:Q3_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"

The first adjusted model I've used that actually seems to be equal-or-better to previous alternatives

by Tribbler - opened May 13

Discussion

Tribbler

May 13

Just wanted to say - good stuff. From my usage it seems equal-or-better to IQ3_XXS while being notably smaller.

Tribbler

May 13

Actually, upon further usage, there's something I did notice - this model seems significantly worse at vision than IQ3_XXS. It starts looping, cutting off output, etc. Might be worth checking out. Also looping propensity in general, possibly as a target for optimization.

deucebucket

Owner May 13

This is something I've noticed too. It's very good at one shot tool calls, and is very good at interaction. So I need to do another ablation review targeting thinking, as that is what seems to cause the most loops. Thanks for trying it out, I hope it's of some use.

deucebucket

Owner May 15

•

edited May 15

Thanks for the report — I investigated this today.

Short version: It is likely quantization damage from v1 being too aggressive, not a C++ bug. v1 put attention QKV and all routed expert weights at Q2_K across all 40 layers, which explains both vision degradation and reasoning looping.

v2 overrides already exist that protect attn_qkv + routed experts at Q3_K_M. I just need to build and test V2
Will update this thread when v2 is available.

Tribbler

May 15

Glad to hear that! Looking forward to it, and thank you for sharing the results of your work.

deucebucket

Owner May 18

Correction after more testing: the vision issue was not confirmed quantization damage in the main v1 GGUF. The original upload was missing the required mmproj-F16.gguf vision projector, so image input was not packaged correctly.

I uploaded the missing mmproj-F16.gguf and updated the README with the proper llama.cpp command:

llama-server \
  --model Qwen3.6-35B-A3B-Cerebellum.gguf \
  --mmproj mmproj-F16.gguf \
  --n-gpu-layers 99 \
  --ctx-size 8192 \
  --reasoning off \
  --reasoning-budget 0

Retested with the projector attached:

Model	Size	Vision smoke	RealWorldQA 200
Cerebellum v1	12 GB	36/36	156/200, 78.0%
Cerebellum v2 test build	15 GB	36/36	156/200, 78.0%
Stock Q3_K_M baseline	16 GB	36/36	155/200, 77.5%

So v1 with the proper vision file performs as well as the stock Q3_K_M baseline on this vision check. The v2 test build added about 3 GB and did not give a worthwhile gain, so I am keeping the corrected v1 package as the recommended release. Thanks again for catching/reporting the vision problem.

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