Instructions to use Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF")
model = AutoModelForMultimodalLM.from_pretrained("Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-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 Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16
# Run inference directly in the terminal:
llama cli -hf Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16
# Run inference directly in the terminal:
llama cli -hf Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

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 Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16
# Run inference directly in the terminal:
./llama-cli -hf Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

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 Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

Use Docker

docker model run hf.co/Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

LM Studio
Jan

vLLM

How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Regic1d3/Qwen3.5-4B-Clarity-1.0-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": "Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

SGLang

How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF with Ollama:
```
ollama run hf.co/Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16
```
Unsloth Desktop

How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

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": "Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF with Docker Model Runner:
```
docker model run hf.co/Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16
```

Lemonade

How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

Run and chat with the model

lemonade run user.Qwen3.5-4B-Clarity-1.0-GGUF-BF16

List all available models

lemonade list

Hermes Agent

How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-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 Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

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 Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16

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 "Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF:BF16" \
  --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"

Qwen3.5-4B — Reasoning Fine-Tune

A lightweight fine-tune of Qwen3.5-4B focused on improving reasoning, attention to detail, and epistemic behavior.

Clarity is the technical name I use for this personal experimental fine-tune. It is not intended to imply a formal benchmark result, standardized capability, or claim of superior model quality.

What changed

Improved logical reasoning on everyday and structured problems
Better multi-step problem solving and consistency checking
Improved attention to contradictions and incorrect assumptions
Some improvement in Python code precision
Additional reasoning examples in English and Russian

The main training data was distilled from Qwen3.8-27B, combined with additional reasoning examples in English and Russian.

This is intended as a relatively subtle behavioral fine-tune rather than a major capability expansion. The goal was to make the model more deliberate and reliable when solving problems, while retaining the general capabilities of the original Qwen3.5-4B.

Note: This is a personal experimental fine-tune. The improvements described above are based primarily on my own qualitative evaluations rather than a standardized benchmark.

Evaluation

Initial qualitative testing suggests improvements in:

constraint-based reasoning
causal reasoning
multi-step arithmetic
error detection and correction
distinguishing facts from assumptions
avoiding invalid logical implications

The model can still produce incorrect reasoning and should not be treated as reliably calibrated. Some tendency toward overconfident or overly elaborate explanations was observed during evaluation.

Sampling parameters

General

Temperature: 0.8
Top-p: 0.95
Top-k: 20
Min-p: 0.05
Repetition penalty: 1.05
Presence penalty: 1.0

Code

Temperature: 0.6
Top-p: 0.95
Top-k: 20
Min-p: 0.0
Repetition penalty: 1.00
Presence penalty: 0.0

tags: - gguf - llama.cpp - llama-cpp - unsloth

Qwen3.5-4B-Clarity-1.0 : GGUF

This model was converted to GGUF format using Unsloth.

Example usage:

For text only LLMs: llama-cli -hf Regic1d3/Qwen3.5-4B-Clarity-1.0 --jinja
For multimodal models: llama-mtmd-cli -hf Regic1d3/Qwen3.5-4B-Clarity-1.0 --jinja

Available model files:

Qwen3.5-4B-Distill-stage3-Inference.Q5_K_M.gguf
Qwen3.5-4B-Distill-stage3-Inference.Q8_0.gguf

Downloads last month: 872

GGUF

Model size

4B params

Architecture

qwen35

Hardware compatibility

5-bit

8-bit

Model tree for Regic1d3/Qwen3.5-4B-Clarity-1.0-GGUF

Base model

Qwen/Qwen3.5-4B-Base

Finetuned

Qwen/Qwen3.5-4B

Quantized

(441)

this model