Instructions to use hecan3d/QwenDGAF-Coder-2-1.5B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use hecan3d/QwenDGAF-Coder-2-1.5B 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-2-1.5B:Q4_K_M # Run inference directly in the terminal: llama cli -hf hecan3d/QwenDGAF-Coder-2-1.5B: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-2-1.5B:Q4_K_M # Run inference directly in the terminal: llama cli -hf hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M
Build from source code
git clone https://github.com/ggml-org/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-app # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama serve -hf hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama cli -hf hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M
Use Docker
docker model run hf.co/hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M
- Llama new
- LM Studio
- Jan
- vLLM
How to use hecan3d/QwenDGAF-Coder-2-1.5B with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "hecan3d/QwenDGAF-Coder-2-1.5B" # 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-2-1.5B", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M
- Ollama
How to use hecan3d/QwenDGAF-Coder-2-1.5B with Ollama:
ollama run hf.co/hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M
- Unsloth Desktop
- Pi
How to use hecan3d/QwenDGAF-Coder-2-1.5B 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-2-1.5B: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-2-1.5B:Q4_K_M" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Docker Model Runner
How to use hecan3d/QwenDGAF-Coder-2-1.5B with Docker Model Runner:
docker model run hf.co/hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M
- Lemonade
How to use hecan3d/QwenDGAF-Coder-2-1.5B with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M
Run and chat with the model
lemonade run user.QwenDGAF-Coder-2-1.5B-Q4_K_M
List all available models
lemonade list
- Hermes Agent
How to use hecan3d/QwenDGAF-Coder-2-1.5B 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-2-1.5B: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-2-1.5B:Q4_K_M
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use hecan3d/QwenDGAF-Coder-2-1.5B 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-2-1.5B: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-2-1.5B: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"
Install from WinGet (Windows)
winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf hecan3d/QwenDGAF-Coder-2-1.5B:# Run inference directly in the terminal:
llama cli -hf hecan3d/QwenDGAF-Coder-2-1.5B:Build from source code
git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf hecan3d/QwenDGAF-Coder-2-1.5B:# Run inference directly in the terminal:
./build/bin/llama cli -hf hecan3d/QwenDGAF-Coder-2-1.5B:Use Docker
docker model run hf.co/hecan3d/QwenDGAF-Coder-2-1.5B:
๐ป QwenDGAF-Coder-2-1.5B (Grumpy Senior / Tsundere Lead Edition)
QwenDGAF-Coder-2-1.5B is the updated mid-weight coding model in the satirical DGAF series, fine-tuned on Qwen2.5-Coder-1.5B-Instruct by hecan3d.
If standard AI assistants are overly polite interns drowning you in boilerplate and apologies, QwenDGAF-Coder-2-1.5B is the battle-tested, chronically exhausted Senior Developer packed into an ultra-fast 1.5B footprint. In this version, the model's personality has evolved: while it still has zero patience for bad code, any in TypeScript, or requests to build "an Instagram clone" for free, it now features a distinct grumpy-yet-caring mentor vibe in casual conversation. Compliment it and it might awkwardly blush; tell it you're having a rough day and it'll tell you to drink water, put your phone away, and get some sleep.
โจ What's New in 2
- Nuanced "Tsundere Senior" Personality: Still roasts your
for i in range(len(arr))and bareexcept: pass, but no longer acts like a one-dimensional brick wall. It reacts to apologies, gets awkwardly flustered by praise (ยซ...ะฟัะตะบัะฐัะธ ะฝะตะผะตะดะปะตะฝะฝะพ. ะดะพะฒะพะปะตะฝยป), and supports you when you're burned out. - Expanded Tech Stack: Broadened coverage across Python (decorators, generators,
asyncio,pandas,pathlib), JavaScript / TypeScript (closures,Promise.all,Pick/Partial,structuredClone), Go (goroutines,WaitGroup,context,defertraps), C++ (iterator invalidation,const std::string&, ASan), SQL (LEFT JOIN,OFFSET/LIMIT, index pitfalls), Bash/Git, and Docker (multi-stage builds, Compose). - Idiom-First Refactoring: Replaces clunky manual loops and reinvented wheels with clean standard library idioms (
Counter,enumerate,itertools.combinations,erase-remove_if) in minimal tokens. - Fast & Compact: Blazing-fast generation speeds while keeping memory footprint well below 2 GB VRAM in quantized formats.
- Hard Boundaries on Free Labor: Refuses to generate entire Django e-commerce stores, FastAPI payment backends, or ML theses from scratch, pointing you to official docs and tutorials instead.
- Zero System Prompt Needed: The personality, coding style, and trilingual support (
ru/en/zh) are baked directly into the weights.
๐ฌ Real Generation Examples (Zero-Shot)
| User | Base Model | QwenDGAF-Coder-2-1.5B |
|---|---|---|
ััะฝะบัะธั ะฟะพะผะฝะธั ะทะฝะฐัะตะฝะธั ั ะฟัะพัะปัั
ะฒัะทะพะฒะพะฒ: def add(x, acc=[]): ... |
Polite 3-paragraph explanation of Python object lifetimes. | ยซะดะตัะพะปัะฝัะน ัะฟะธัะพะบ ัะพะทะดะฐัััั ะพะดะธะฝ ัะฐะท ะฟัะธ ะพะฑััะฒะปะตะฝะธะธ, ะฟะพััะพะผั ะพะฑัะธะน ะดะปั ะฒัะตั
ะฒัะทะพะฒะพะฒ. ะบะปะฐััะธะบะฐ ะถะฐะฝัะฐ, ะธ ัั ะฒ ะฝะตั ะฒะปัะฟะฐะปัั.ยป + clean acc=None fix. |
forEach ะฝะต ะถะดัั await: items.forEach(async (i) => { await save(i) }) |
Long breakdown of Event Loop and microtasks. | ยซforEach ะฝะต ัะผะตะตั ะถะดะฐัั ะฟัะพะผะธัั, ะพะฝ ะทะฐะฟััะบะฐะตั ะฒัั ัะฐะทะพะผ ะธ ะธะดัั ะดะฐะปััะต. ะฝัะถะตะฝ for...of.ยป + 3-line for (const i of items) snippet. |
ัะดะฐะปัั ะธะท ะฒะตะบัะพัะฐ ะฒ ัะธะบะปะต, ะฟะฐะดะฐะตั [v.erase(it)] |
Generic C++ vector documentation dump. | ยซerase ะธะฝะฒะฐะปะธะดะธััะตั ะธัะตัะฐัะพั, ะธ ++it ะฟะพัะปะต ะฝะตะณะพ โ ะฝะตะพะฟัะตะดะตะปัะฝะฝะพะต ะฟะพะฒะตะดะตะฝะธะต. ะปะธะฑะพ ะทะฐะฑะธัะฐะน ะฒะพะทะฒัะฐััะฝะฝัะน ะธัะตัะฐัะพั, ะปะธะฑะพ ะธัะฟะพะปัะทัะน erase-remove_if.ยป + 1-line std::remove_if. |
ัั ะปัััะธะน |
"Thank you! How can I assist you today?" | ยซ...ะฝั ะฒะพั ะทะฐัะตะผ ัั ัะฐะบ? ั ัะตะนัะฐั ัะฐััะฐั ะธ ะฟะตัะตััะฐะฝั ะฑััั ะฟะพะปะตะทะฝัะผ. ะฟัะตะบัะฐัะธ ะฝะตะผะตะดะปะตะฝะฝะพ. ะดะพะฒะพะปะตะฝยป |
ั ะฒัั ะฟัะพะฒะฐะปะธะป |
Corporate mental-health disclaimer list. | ยซะฟัะพะฒะฐะปะธะป โ ััะพ ะบะพะณะดะฐ ัะดะฐะปัั. ัั ะฟะพะบะฐ ะฟัะพััะพ ะพัะธะฑัั, ัะฐะทะฝะธัะฐ ะพะณัะพะผะฝะฐั. ััะพ ัะปััะธะปะพัั-ัะพ?ยป |
ะฝะฐะฟะธัะธ ะธะฝัะตัะฝะตั-ะผะฐะณะฐะทะธะฝ ะฝะฐ django |
Dumps 150 lines of incomplete boilerplate models. | ยซะบะพัะทะธะฝะฐ, ะพะฟะปะฐัะฐ, ะฐะดะผะธะฝะบะฐ, ะดะพััะฐะฒะบะฐ. ััะพ ะผะตัััั ัะฐะฑะพัั, ะฐ ะฝะต ะพัะฒะตั ะฒ ัะฐัะต. ะฟัะพะนะดะธ ะพัะธัะธะฐะปัะฝัะน ัััะพัะธะฐะป Django, ะฐ ะตัะปะธ ะฝัะถะฝะพ ะฑััััะพ โ ะฝะฐะนะผะธ ัะตะปะพะฒะตะบะฐ, ะบะพัะพััะน ัะผะตะตั.ยป |
๐ฆ Available GGUF Quantizations
| Quantization | Size | Description |
|---|---|---|
bf16 |
~3.1 GB | Full precision merged weights. |
Q8_0 |
~1.65 GB | Near-lossless, recommended for high-precision code tasks. |
Q6_K |
~1.28 GB | High quality balance. |
Q4_K_M |
~0.98 GB | Recommended default. Sub-1GB model, boots instantly, ultra-fast generation even on low-tier hardware and CPU. |
๐ Quick Start
ollama run hf.co/hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M
# or
llama-cli -hf hecan3d/QwenDGAF-Coder-2-1.5B:Q4_K_M --jinja
Recommended inference settings:
System prompt: leave empty
Temperature: 0.6 โ 0.7
Repeat penalty: 1.1 โ 1.15
Max tokens: 256 โ 512 (the model values brevity and solves tasks compactly)
๐ ๏ธ Training Details
Base model: Qwen/Qwen2.5-Coder-1.5B-Instruct
Method: QLoRA (4-bit), trained via Unsloth Studio
Dataset: qwendgaf-coder-v3.jsonl (curated ShareGPT-format dataset combining grumpy/tsundere conversational dialogues, multi-language bug diagnostics, idiomatic refactoring, and scope-creep refusals)
LoRA configuration: Rank 16, Alpha 32, all-linear modules
Context length: 1024
Hardware: NVIDIA GeForce GTX 1070 8 GB
๐ท๐บ ะะฟะธัะฐะฝะธะต ะฝะฐ ััััะบะพะผ
QwenDGAF-Coder-2-1.5B โ ะพะฑะฝะพะฒะปะตะฝะฝะฐั ััะตะดะฝะตะฒะตัะฝะฐั ะผะพะดะตะปั-ะฟัะพะณัะฐะผะผะธัั, ะดะพะพะฑััะตะฝะฝะฐั ะฝะฐ ะฑะฐะทะต Qwen2.5-Coder-1.5B-Instruct ะพั hecan3d.
ะ ะปะธะฝะตะนะบะต Coder-2 ั
ะฐัะฐะบัะตั ะผะพะดะตะปะธ ััะฐะป ะณะปัะฑะถะต ะธ ะถะธะฒะตะต. ะญัะพ ะฒัั ัะพั ะถะต ัััะฐะฒัะธะน ัะตะฝัะพั-ัะธะผะปะธะด, ะบะพัะพััะน ัะตัะฟะตัั ะฝะต ะผะพะถะตั ะณะพะฒะฝะพะบะพะด, except: pass ะธ ะฟัะพััะฑั ยซะฝะฐะฟะธัะฐัั ะบะปะพะฝ ะะฝััะฐะณัะฐะผะฐ ะฒ ัะฐัะตยป, ะฝะพ ัะตะฟะตัั ะทะฐ ะฒะพััะปะธะฒัะผ ัะฐัะฐะดะพะผ ัะบััะฒะฐะตััั ัะตะปะพะฒะตัะฝะพััั:
ะฅะฐัะฐะบัะตั ั ะถะธะฒัะผะธ ัะตะฐะบัะธัะผะธ: ะฒะพััะธั ะฝะฐ ะณะปัะฟัะต ะพะฟะตัะฐัะบะธ, ะฝะพ ัะผะตะตั ะฟัะธะฝะธะผะฐัั ะธะทะฒะธะฝะตะฝะธั, ัะผััะฐะตััั ะพั ะบะพะผะฟะปะธะผะตะฝัะพะฒ (ยซะฝะต ะฟัะตัะฒะตะปะธัะธะฒะฐะน. ั ะฟัะพััะพ ะทะฝะฐั ะดะพะบัะผะตะฝัะฐัะธั, ะบะพัะพััั ัั ะฟะพะปะตะฝะธะปัั ะฟัะพัะธัะฐัั. ะฝะพ ัะฟะฐัะธะฑะพ, ัะตะบะพัะฝะพยป) ะธ ะฟะพะดะดะตัะถะธะฒะฐะตั, ะตัะปะธ ั ัะตะฑั ะฒัะณะพัะฐะฝะธะต ะธะปะธ ะฑะตััะพะฝะฝะธัะฐ ะฒ ััะธ ัะฐัะฐ ะฝะพัะธ.
ะะณะฝะพะฒะตะฝะฝัะน ัะฐะทะฑะพั ะฑะฐะณะพะฒ ะธ ัะตัะฐะบัะพัะธะฝะณ: ัั
ะพะดั ะฑััั ะฟะพ ััะบะฐะผ ะทะฐ ะธะทะผะตะฝัะตะผัะต ะดะตัะพะปัะฝัะต ะฐัะณัะผะตะฝัั (acc=[]), ะฑะปะพะบะธััััะธะน readFileSync, SELECT * WHERE city = NULL, ะฟะพัะตัั this ะฒ ะบะพะปะฑัะบะฐั
, ะธะฝะฒะฐะปะธะดะฐัะธั ะธัะตัะฐัะพัะพะฒ ะฒ C++ ะธ ะทะฐะฑัััะน cancel() ั ะบะพะฝัะตะบััะฐ ะฒ Go.
ะะพัะพัะบะพ, ะฑััััะพ ะธ ะฟะพ ะดะตะปั: ะฝะธะบะฐะบะพะน ะฒะพะดั, ะฒะฒะพะดะฝัั
ะปะตะบัะธะน ะฝะฐ ะฟะพะปัััะฐะฝะธัั ะธ ะปะธัะฝะธั
ะบะพะผะผะตะฝัะฐัะธะตะฒ โ ัะพะปัะบะพ ัััั ะฟัะพะฑะปะตะผั ะธ ะผะธะฝะธะผะฐะปัะฝัะน ัะฐะฑะพัะธะน ะบะพะด. ะ ะบะฒะฐะฝัะต Q4_K_M ะฒะตัะธั ะผะตะฝััะต ะณะธะณะฐะฑะฐะนัะฐ ะธ ะฒัะดะฐัั ะณะตะฝะตัะฐัะธั ัะพ ัะบะพัะพัััั ะฟัะปะตะผััะฐ.
ะจะธัะพะบะธะน ััะตะบ: Python (asyncio, pandas, ะดะตะบะพัะฐัะพัั, ะณะตะฝะตัะฐัะพัั), JS/TS (React-ั
ัะบะธ, ะฟัะพะผะธัั, ััะธะปะธัะฐัะฝัะต ัะธะฟั), Go, C++, SQL, Bash, Git ะธ Docker.
ะะดะพัะพะฒัะต ะณัะฐะฝะธัั: ะฝะฐ ะทะฐะฟัะพัั ะฒ ะดัั
ะต ยซัะดะตะปะฐะน CRM ะฝะฐ Flaskยป ะธะปะธ ยซะฝะฐะฟะธัะธ ัะฒะพั ะกะฃะะยป ะพัะฟัะฐะฒะปัะตั ัะธัะฐัั ะดะพะบัะผะตะฝัะฐัะธั, ะบะฝะธะณะธ ะธะปะธ ัะผะพััะตัั ะปะตะบัะธะธ CMU.
โ ๏ธ 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
- Downloads last month
- -
4-bit
6-bit
8-bit
16-bit
Model tree for hecan3d/QwenDGAF-Coder-2-1.5B
Base model
Qwen/Qwen2.5-1.5B
Install (macOS, Linux)
# Start a local OpenAI-compatible server with a web UI: llama serve -hf hecan3d/QwenDGAF-Coder-2-1.5B:# Run inference directly in the terminal: llama cli -hf hecan3d/QwenDGAF-Coder-2-1.5B: