README.md

metadata

title: 𝐂𝐡𝐚𝐭 𝐇𝐮𝐦𝐨𝐮𝐫 𝐋𝐋𝐌 𝟎.𝟓𝐁
emoji: 🐨
colorFrom: purple
colorTo: yellow
sdk: docker
pinned: false

🧠 DNAI Humour Chatbot - Interactive Web Interface

A beautiful, fully-functional chat interface for the dnai-humour-0.5B-instruct model. Chat with a witty, lightweight AI assistant that's fast, friendly, and surprisingly capable.

---

✨ Features

• 💬 Real-time Chat - Smooth, responsive conversation interface
• 🎨 Beautiful UI - Modern design with dark/light mode
• ⚡ Fast Responses - Optimized for 0.5B parameter model
• 🎭 Personality - Witty and helpful, not robotic
• 📱 Responsive - Works on all devices
• 🧹 Clean UX - Clear conversations, suggestions, timestamps

---

🚀 Quick Start

Prerequisites

• Python 3.8+
• CUDA-capable GPU (recommended) or CPU
• 2GB+ VRAM (GPU) or 4GB+ RAM (CPU)

Installation

1. Clone the Space

git clone https://huggingface.co/spaces/YOUR-USERNAME/dnai-humour-chatbot
cd dnai-humour-chatbot

2. Install Dependencies

pip install -r requirements.txt

3. Run the Application

uvicorn app:app --host 0.0.0.0 --port 7860

4. Open Browser

http://localhost:7860

---

🌐 Deploy to Hugging Face Spaces

Step 1: Create Space

1. Go to Hugging Face
2. Click New Space
3. Settings:
   • Name: dnai-humour-chatbot
   • SDK: Docker
   • Hardware: T4 Small (or CPU basic for testing)
   • Visibility: Public

Step 2: Upload Files

Upload these files to your Space:

dnai-humour-chatbot/
├── app.py
├── requirements.txt
├── index.html
├── README.md
└── Dockerfile (optional)

Step 3: Create Dockerfile (if needed)

FROM python:3.10-slim

WORKDIR /app

# Install dependencies
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

# Copy application
COPY . .

# Expose port
EXPOSE 7860

# Run
CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

Step 4: Wait for Build

• Hugging Face will automatically build your Space
• Check logs for any errors
• Model will download on first startup (~500MB)
• Once "Running", your chatbot is live! 🎉

---

📁 File Structure

dnai-humour-chatbot/
│
├── app.py                    # FastAPI backend with model inference
│   ├── /api/chat            # POST - Send messages
│   ├── /api/info            # GET - Model information
│   ├── /health              # GET - Health check
│   └── /api/reset           # POST - Reset conversation
│
├── index.html               # React-based chat interface
│   ├── Message history
│   ├── Dark/Light mode
│   ├── Typing indicators
│   └── Quick suggestions
│
├── requirements.txt         # Python dependencies
│   ├── transformers
│   ├── torch
│   ├── fastapi
│   └── uvicorn
│
└── README.md               # This file

---

🔌 API Documentation

POST /api/chat

Request:

{
  "messages": [
    {
      "role": "user",
      "content": "Tell me a joke"
    }
  ],
  "temperature": 0.7,
  "max_tokens": 256
}

Response:

{
  "response": "Why don't scientists trust atoms? Because they make up everything! 😄",
  "model": "DarkNeuron-AI/dnai-humour-0.5B-instruct",
  "tokens_used": 45
}

GET /api/info

Response:

{
  "model_name": "DNAI Humour 0.5B Instruct",
  "version": "1.0",
  "base_model": "Qwen2.5-0.5B-Instruct",
  "parameters": "~0.5 Billion",
  "capabilities": [
    "Instruction following",
    "Conversational AI",
    "Light humor",
    "Low-latency responses"
  ]
}

---

🎨 UI Features

Chat Interface

• Clean, modern design
• Message bubbles with timestamps
• User/Assistant avatars
• Smooth animations

Dark/Light Mode

• Toggle between themes
• Smooth transitions
• Persistent preferences (client-side)

Smart Suggestions

• Quick-start prompts
• Contextual examples
• One-click input

Typing Indicators

• Real-time feedback
• Loading animations
• Response timing

---

⚙️ Configuration

Model Parameters (app.py)

MODEL_NAME = "DarkNeuron-AI/dnai-humour-0.5B-instruct"
MAX_LENGTH = 512          # Context window
TEMPERATURE = 0.7         # Creativity (0.0-1.0)
TOP_P = 0.9              # Nucleus sampling
TOP_K = 50               # Top-k sampling

Generation Settings

Adjust in /api/chat request:

• temperature: 0.1 (focused) to 1.0 (creative)
• max_tokens: 50 (short) to 512 (long)
• stream: true/false (streaming support)

---

🐛 Troubleshooting

Issue: Model not loading

Symptoms: 503 errors, "Model not loaded"

Solutions:

# Check CUDA availability
python -c "import torch; print(torch.cuda.is_available())"

# Verify model download
python -c "from transformers import AutoModelForCausalLM; AutoModelForCausalLM.from_pretrained('DarkNeuron-AI/dnai-humour-0.5B-instruct')"

# Check logs
uvicorn app:app --log-level debug

Issue: Out of memory

Solutions:

1. Reduce max_tokens in generation
2. Use CPU instead of GPU (slower but works)
3. Enable model quantization (INT8)

# In app.py, modify model loading:
model = AutoModelForCausalLM.from_pretrained(
    MODEL_NAME,
    load_in_8bit=True,  # Enable INT8 quantization
    device_map="auto"
)

Issue: Slow responses

Solutions:

1. Use GPU instead of CPU
2. Reduce max_tokens
3. Lower temperature for faster sampling
4. Use smaller batch size

---

📊 Performance Benchmarks

Hardware	Response Time	Memory Usage
T4 GPU	~1-2s	~2GB VRAM
CPU	~5-10s	~4GB RAM
A10G GPU	~0.5-1s	~2GB VRAM

---

🎯 Use Cases

• Educational Chatbots - Learning companions
• Personal Assistants - Quick help & info
• Code Helpers - Programming Q&A
• Creative Writing - Brainstorming & ideas
• General Chat - Friendly conversation

---

🚫 Limitations

• Not for production medical/legal advice
• Limited context window (512 tokens)
• 0.5B parameters - expect occasional mistakes
• No long-term memory - each conversation is independent
• English-focused - other languages may be limited

---

🔒 Privacy & Safety

• No data logging - conversations not stored
• Local processing - your data stays with you
• No tracking - no analytics or monitoring
• Open source - fully transparent code

---

🛠️ Advanced Customization

Change Model Personality

Edit the system prompt in app.py:

def format_chat_prompt(messages: List[Message]) -> str:
    system_prompt = "You are a helpful, witty AI assistant."
    formatted_messages = [f"System: {system_prompt}"]
    # ... rest of code

Add Memory

Implement conversation history storage:

# Simple in-memory storage
conversations = {}

@app.post("/api/chat")
async def chat(request: ChatRequest, user_id: str = "default"):
    if user_id not in conversations:
        conversations[user_id] = []
    # ... use stored history

Enable Streaming

For real-time token-by-token responses:

from fastapi.responses import StreamingResponse

@app.post("/api/chat/stream")
async def chat_stream(request: ChatRequest):
    def generate():
        # Yield tokens as they're generated
        for token in model.generate_stream(...):
            yield f"data: {token}\n\n"
    
    return StreamingResponse(generate(), media_type="text/event-stream")

---

📞 Support

For issues, questions, or suggestions:

• Open an issue on GitHub
• Contact via Hugging Face
• Check model card: DarkNeuron-AI/dnai-humour-0.5B-instruct

---

🙏 Acknowledgments

• Base Model: Qwen2.5-0.5B-Instruct by Alibaba
• Dataset: OpenAssistant v1
• Framework: Hugging Face Transformers
• UI: React + Tailwind CSS

---

📝 License

MIT License - Free to use, modify, and distribute

---

Crafted with ❤️ and passion by @MADARA369Uchiha

Small brain, well-trained. Fast responses, good vibes. ✨