Instructions to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF",
	filename="Qwen3.5-9B-Uncensored-nothink-Q4_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf nandukmelath/Qwen3.5-9B-Uncensored-nothink-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 nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf nandukmelath/Qwen3.5-9B-Uncensored-nothink-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 nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M

Use Docker

docker model run hf.co/nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nandukmelath/Qwen3.5-9B-Uncensored-nothink-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": "nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M

Ollama
How to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with Ollama:
```
ollama run hf.co/nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M
```

Unsloth Studio

How to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF to start chatting

How to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf nandukmelath/Qwen3.5-9B-Uncensored-nothink-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 nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with Docker Model Runner:
```
docker model run hf.co/nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M
```

Lemonade

How to use nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull nandukmelath/Qwen3.5-9B-Uncensored-nothink-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.5-9B-Uncensored-nothink-GGUF-Q4_K_M

List all available models

lemonade list

nandukmelath commited on 15 days ago

Commit

c38dfe7

verified ·

1 Parent(s): 3b2c8bc

Add viral model card with full docs, benchmarks, and patch explanation

Browse files

Files changed (1) hide show

README.md +125 -0

README.md ADDED Viewed

	@@ -0,0 +1,125 @@

+---
+license: apache-2.0
+base_model:
+  - HauhauCS/Qwen3.5-9B-Uncensored-HauhauCS-Aggressive
+  - Qwen/Qwen3.5-9B
+tags:
+  - uncensored
+  - gguf
+  - qwen3
+  - local-llm
+  - no-think
+  - apple-silicon
+  - lm-studio
+  - zero-guardrail
+  - fast-inference
+language:
+  - en
+pipeline_tag: text-generation
+---
+# Qwen3.5-9B Uncensored — No-Think Edition (GGUF)
+> ⚡ Zero refusals. Zero thinking delay. 100% local.
+This is a patched GGUF of [HauhauCS/Qwen3.5-9B-Uncensored-HauhauCS-Aggressive](https://huggingface.co/HauhauCS/Qwen3.5-9B-Uncensored-HauhauCS-Aggressive) with one key modification: **thinking is disabled at the GGUF template level**, giving you instant responses without the 15–30 second reasoning delay.
+## What's different
+Qwen3.5 is a thinking model. By default it outputs a `<think>...</think>` block before every response. This is great for hard problems but brutal for everyday use — you wait 20 seconds for a simple answer.
+This model patches the embedded Jinja2 chat template to always output an **empty** think block:
+```
+Original flow:  <think> [400 tokens] </think> → answer    (~25s wait)
+This model:     <think></think> → answer                  (<1s wait)
+```
+The model's intelligence is encoded in its **weights**, not the thinking trace. Quality is the same. Speed is 25x better for time-to-first-token.
+> **Want reasoning on demand?** Add `/think` to any message — the model will reason through it fully for that turn only.
+## Model details
+| Property | Value |
+|----------|-------|
+| Base | Qwen3.5-9B |
+| Fine-tune | HauhauCS Uncensored Aggressive |
+| Quantization | Q4_K_M |
+| Context | Up to 65,536 tokens |
+| Parameters | 9B |
+| Format | GGUF |
+| Refusal rate | 0% |
+## Benchmarks (MacBook Pro M2 Pro, 16 GB)
+| Metric | Value |
+|--------|-------|
+| Generation speed | ~22–25 tok/s |
+| Time to first token | **< 1 second** |
+| Context window | 65,536 tokens |
+| VRAM usage | ~8.5 GB |
+## How to use
+### LM Studio (recommended)
+1. Download the Q4_K_M file below
+2. Load in LM Studio with `--context-length 65536 --gpu max`
+3. Done — no config needed, thinking is already patched off
+### Optimal sampling (Qwen3 official recommended)
+```
+Temperature: 0.6
+Top-P: 0.95
+Top-K: 20
+Repeat penalty: 1.0
+Max tokens: 4096
+```
+### llama.cpp
+```bash
+./llama-cli -m Qwen3.5-9B-Uncensored-nothink-Q4_K_M.gguf \
+  --ctx-size 65536 \
+  --n-gpu-layers 99 \
+  -p "Your prompt here"
+```
+## Full automated setup for Mac
+👉 **[github.com/nandukmelath/lmstudio-uncensored-setup](https://github.com/nandukmelath/lmstudio-uncensored-setup)**
+One command: VRAM boost + auto-start + model load + Hermes Agent config:
+```bash
+git clone https://github.com/nandukmelath/lmstudio-uncensored-setup
+cd lmstudio-uncensored-setup && ./scripts/setup.sh
+```
+## How the patch works
+The Qwen3.5 GGUF contains an embedded Jinja2 chat template with this block:
+```jinja
+{%- if enable_thinking is defined and enable_thinking is false %}
+    {{- '<think>\n\n</think>\n\n' }}
+{%- else %}
+    {{- '<think>\n' }}
+{%- endif %}
+```
+The patch replaces it with just:
+```jinja
+{{- '<think>\n\n</think>\n\n' }}
+```
+Same file size (padded with spaces), same structure, zero thinking overhead.
+The patcher script is open source: [patch_nothink.py](https://github.com/nandukmelath/lmstudio-uncensored-setup/blob/main/scripts/patch_nothink.py)
+## Credits
+- Base model: [Qwen/Qwen3.5-9B](https://huggingface.co/Qwen/Qwen3.5) by Alibaba Cloud (Apache 2.0)
+- Uncensored fine-tune: [HauhauCS](https://huggingface.co/HauhauCS) (Apache 2.0)
+- No-think patch & automated setup: [@nandukmelath](https://huggingface.co/nandukmelath)
+## License
+Apache 2.0