neuron-v0.4 0.4.91

A local AI assistant with OpenLLaMA and Mistral models, advanced identity protection and hardware optimization

Neuron AI Assistant

A powerful local AI assistant with advanced identity protection, hardware optimization, and comprehensive conversation management.

Created by: Dev Patel
Version: 0.4.91

Features

• Identity Protection: Built-in safeguards against prompt injection and identity tampering
• Hardware Optimization: Auto-detects CPU, GPU (CUDA), Apple Silicon (MPS), RAM, and VRAM
• Multiple Models: Support for OpenLLaMA 7B v2 (CPU/GPU) and Mistral 7B (GPU-optimized)
• Conversation Management: Save, export, and manage chat history
• Config Security: Cryptographic signing and automatic backups
• Error Recovery: Automatic backup restoration and config migration
• Resource Management: Dynamic token limits and OOM handling
• Diagnostic Tools: Built-in system health checks

Requirements

• Python: 3.8 or higher
• RAM: Minimum 4GB (8GB+ recommended)
• Disk Space: 20GB free (for model downloads)
• GPU (Optional): NVIDIA with CUDA support for better performance

Installation

Option 1: From PyPI (when published)

pip install neuron-ai-assistant

Option 2: From Source

# Clone the repository
git clone https://github.com/devpatel/neuron-ai-assistant.git
cd neuron-ai-assistant

# Install dependencies
pip install -r requirements.txt

# Or install with all features
pip install -e .[all]

Option 3: GPU Support

# For NVIDIA GPU (CUDA 11.8)
pip install torch==2.0.0+cu118 --extra-index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt

# For NVIDIA GPU (CUDA 12.1)
pip install torch==2.0.0+cu121 --extra-index-url https://download.pytorch.org/whl/cu121
pip install -r requirements.txt

Option 4: CPU Only (Smaller)

pip install torch==2.0.0+cpu --extra-index-url https://download.pytorch.org/whl/cpu
pip install -r requirements.txt

Quick Start

First Run

python neuron_assistant.py

On first run, you'll be asked to:

1. Enter your name
2. Select a model (OpenLLaMA 7B v2 or Mistral 7B)
3. Wait for model download (if needed)

Using the Assistant

# After installation
neuron
# or
neuron-assistant

💻 Commands

Command	Description
/help	Show available commands
/clear	Clear conversation history
/save	Save conversation to text file
/export	Export conversation to JSON
/stats	Show system statistics
/tokens <n>	Set max tokens (16-1024)
/model	Change AI model
/migrate	Fix/update old configs
/diagnose	Run system diagnostics
/reset	Reset assistant completely
/exit	Exit gracefully

🔧 Configuration

The assistant creates these files automatically:

• config.json - User and model settings
• config.sig - Cryptographic signature
• models/ - Downloaded AI models
• backups/ - Config backups (last 5)
• .neuron.lock - Instance lock file

Model Comparison

Model	Size	RAM	VRAM	Speed	Quality
OpenLLaMA 7B v2	13.5GB	8GB	0GB	Medium	Excellent
Mistral 7B	14GB	16GB	12GB	Medium	Excellent

Advanced Usage

Set HuggingFace Token

export HF_TOKEN="your_token_here"
python neuron_assistant.py

Custom Token Limit

from neuron_assistant import NeuronAssistant

assistant = NeuronAssistant()
assistant.set_max_tokens(256)

Programmatic Use

from neuron_assistant import NeuronAssistant

# Initialize
assistant = NeuronAssistant(hf_token="optional_token")

# Chat
response = assistant.chat("Hello! How are you?")
print(response)

# Save conversation
assistant.save_history("my_chat.txt")
assistant.export_history_json("my_chat.json")

🐛 Troubleshooting

Model Download Fails

# Check disk space
df -h

# Verify internet connection
ping huggingface.co

# Manual download location
ls models/

Config Corrupted

# Run diagnostics
# In chat: /diagnose

# Migrate config
# In chat: /migrate

# Last resort - reset
# In chat: /reset

Out of Memory

# Use OpenLLaMA model instead of Mistral
# Reduce token limit: /tokens 64
# Clear history: /clear

GPU Not Detected

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

# Reinstall PyTorch with CUDA
pip install torch==2.0.0+cu118 --extra-index-url https://download.pytorch.org/whl/cu118

🔒 Security Features

• Creator Lock: Hardcoded creator name prevents identity theft
• Config Signing: RSA signatures verify config integrity
• Prompt Injection Detection: Blocks manipulation attempts
• Output Sanitization: Removes references to other AI companies
• Backup System: Auto-backups before changes

License

MIT License - See LICENSE file for details

Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Submit a pull request

Support

• Issues: GitHub Issues
• Email: [e-mail]
• Docs: Wiki

Acknowledgments

• Built with PyTorch
• Uses Transformers
• Supports OpenLLaMA
• Model: Mistral AI

Changelog

v0.4.91 (Current)

• OpenLLaMA 7B v2 and Mistral 7B model support
• Removed GPT4All dependency
• Enhanced model selection with smart recommendations
• Improved first-run experience
• Advanced identity protection
• Config migration system
• Comprehensive diagnostics
• Improved error handling
• Backup/restore functionality

---

Made with heart by Dev Patel