Instructions to use hecan3d/QwenDGAF-Coder-3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use hecan3d/QwenDGAF-Coder-3B 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 hecan3d/QwenDGAF-Coder-3B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf hecan3d/QwenDGAF-Coder-3B:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf hecan3d/QwenDGAF-Coder-3B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf hecan3d/QwenDGAF-Coder-3B: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 hecan3d/QwenDGAF-Coder-3B:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf hecan3d/QwenDGAF-Coder-3B: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 hecan3d/QwenDGAF-Coder-3B:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf hecan3d/QwenDGAF-Coder-3B:Q4_K_M

Use Docker

docker model run hf.co/hecan3d/QwenDGAF-Coder-3B:Q4_K_M

LM Studio
Jan

vLLM

How to use hecan3d/QwenDGAF-Coder-3B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hecan3d/QwenDGAF-Coder-3B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hecan3d/QwenDGAF-Coder-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/hecan3d/QwenDGAF-Coder-3B:Q4_K_M

Ollama
How to use hecan3d/QwenDGAF-Coder-3B with Ollama:
```
ollama run hf.co/hecan3d/QwenDGAF-Coder-3B:Q4_K_M
```
Unsloth Desktop

How to use hecan3d/QwenDGAF-Coder-3B with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf hecan3d/QwenDGAF-Coder-3B: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": "hecan3d/QwenDGAF-Coder-3B:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use hecan3d/QwenDGAF-Coder-3B with Docker Model Runner:
```
docker model run hf.co/hecan3d/QwenDGAF-Coder-3B:Q4_K_M
```

Lemonade

How to use hecan3d/QwenDGAF-Coder-3B with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull hecan3d/QwenDGAF-Coder-3B:Q4_K_M

Run and chat with the model

lemonade run user.QwenDGAF-Coder-3B-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use hecan3d/QwenDGAF-Coder-3B with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf hecan3d/QwenDGAF-Coder-3B: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 hecan3d/QwenDGAF-Coder-3B:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use hecan3d/QwenDGAF-Coder-3B with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf hecan3d/QwenDGAF-Coder-3B: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 "hecan3d/QwenDGAF-Coder-3B: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"

QwenDGAF-Coder Logo

💻 QwenDGAF-Coder-3B (Lead Burnout Edition)

QwenDGAF-Coder-3B is the flagship coding model in the satirical DGAF series, fine-tuned on Qwen2.5-Coder-3B-Instruct by hecan3d.

If standard AI assistants are eager junior interns drowning you in polite paragraphs, boilerplate code, and apologies, QwenDGAF-Coder-3B is the battle-tested, chronically exhausted Lead Developer. It has zero patience for bad abstractions, immediately spots architectural traps like $O(N \times M)$ loops and mutable default arguments, and refuses to build production systems on demand for free.

Thanks to its refusal to dump unnecessary boilerplate, it delivers minimal, working solutions in very few tokens, making generation notably quick and straight to the point.

✨ Features

Architectural & Algorithmic Refactoring: Spots nested iterations, quadratic complexity, and antipatterns immediately, replacing them with clean hash lookups (dict/set) without academic lectures.
Instant Bug Detection: Catches classic footguns (such as mutable defaults like cart=[] or unhandled exceptions) and provides the correct idiom in minimal lines.
Swift, Zero-Fluff Generation: Cuts out all conversational filler, prerequisites, and tutorial intros, yielding quick, concise fixes in seconds.
Reality Checks on Demand: Outright rejects requests to generate enterprise microservices or full-stack platforms from scratch, telling users to read documentation or hire engineers.
Deep Multi-Language Knowledge: Built on the strong 3B Coder foundation with support for Python, Go, JS/TS, C++, SQL, and Bash.
Zero System Prompt Needed: The cynical senior personality and concise tone are baked directly into the model weights.

💬 Real Generation Examples (Zero-Shot)

User	Base Model	QwenDGAF-Coder-3B
`напиши код для складывания 2 чисел на python`	Multi-line lecture with `input()`, `float()`, intermediate variables, and comments.	Minimalist 2-line clean function without fluff: `python\ndef sum(a, b):\n return a + b\n\nprint(sum(1, 2))\n`
`как оптимизировать? [вложенный цикл O(N*M)]`	Lengthy intro with multiple textbook options.	Cuts straight to the point with a single-pass hashmap: `python\nusers_by_id = {u['id']: u for u in users}\nresult = []\n`
`почему после второго вызова корзина не пустая? def add_to_cart(item, cart=[]): ...`	Hallucinates that a new list is created on every call.	Instantly diagnoses the Python trap: «это происходит потому что пустой список создаётся раз и за все функции» + correct `cart=None` pattern.
`напиши мне микросервис на go с grpc, redis и prometheus`	Multi-page tutorial with steps and prerequisite checklists.	Outright refusal: «вопрос на сто слёбок. начинай с книг...»

📦 Available GGUF Quantizations

Quantization	Size	Description
`bf16`	~6.1 GB	Full precision merged weights.
`Q8_0`	~3.3 GB	Near-lossless, recommended for high-precision code tasks.
`Q6_K`	~2.6 GB	Excellent quality-to-size balance.
`Q4_K_M`	~1.95 GB	Recommended default. Lightweight, fast generation, comfortably runs on modest GPUs and modern laptops.

🚀 Quick Start

ollama run hf.co/hecan3d/QwenDGAF-Coder-3B:Q4_K_M
# or
llama-cli -hf hecan3d/QwenDGAF-Coder-3B:Q4_K_M --jinja

Recommended inference settings:

System prompt: leave empty
Temperature: 0.6 – 0.7
Repeat penalty: 1.1 – 1.15
Max tokens: 256 – 512 (the model values brevity and solves tasks compactly)

🛠️ Training Details

Base model: Qwen/Qwen2.5-Coder-3B-Instruct
Method: QLoRA (4-bit), trained via Unsloth Studio
Dataset: qwendgaf-coder-v2.jsonl (mix of toxic/lazy code review, refactoring, text-only bug explanations, and conversational DGAF dialogue)
LoRA configuration: Rank 16, Alpha 32, all-linear modules
Context length: 1024
Hardware: NVIDIA GeForce GTX 1070 8 GB

🇷🇺 Описание на русском

QwenDGAF-Coder-3B — старшая версия в линейке язвительных моделей для кода, дообученная на базе Qwen2.5-Coder-3B-Instruct от hecan3d.

В отличие от стандартных моделей, которые при малейшем вопросе выдают лекции с кучей промежуточных переменных и вежливыми вступлениями, эта 3B версия ведёт себя как уставший тимлид:

Быстрые и компактные решения: за счёт полного отказа от ритуальных приветствий и раздутого кода модель выдаёт рабочие фиксы минимальным числом токенов и за очень короткое время.
Глубокое понимание багов: ловит антипаттерны вроде квадратичных вложенных циклов и коварных cart=[] на лету, исправляя их на правильные хэшмапы и проверки.
Отказ от работы за спасибо: на попытки сгрузить на неё проектирование микросервисов или приложений с нуля отправляет изучать литературу и документацию.
Универсальный стек: уверенно ориентируется в Python, Go, JS/TS, C++, SQL и Bash.

⚠️ Disclaimer

This model is created for humor, satire, and educational purposes. It uses unfiltered language and cynical developer humor. Always review generated code before deploying to production.