Local LLM Setup with DeepSeek and Ollama 🚀

A comprehensive toolkit for setting up and optimizing DeepSeek models locally using Ollama. Designed for both human developers and agentic IDEs (like Cursor, Cline, or Windsurf).

Quick Start 🏃

# Clone this repository
git clone https://github.com/ccross2/llocal-llm-startup.git
cd llocal-llm-startup

# Run the automated setup script
chmod +x setup.sh
sudo ./setup.sh

Table of Contents 📑

1. Features
2. Prerequisites
3. Installation
4. Scripts Overview
5. Configuration
6. Usage
7. Optimization
8. Monitoring
9. Troubleshooting

Features ✨

• Automated system analysis and model selection
• Smart optimization for both desktop and laptop environments
• Real-time performance monitoring
• Thermal and power management for laptops
• Benchmark testing suite
• Agentic IDE integration support

Prerequisites 📋

Required Software

# Update package list
sudo apt update

# Install Python 3.10+ and development tools
sudo apt install -y python3.10 python3.10-venv python3-pip build-essential

# Install system monitoring tools
sudo apt install -y htop nvidia-smi sensors

Python Dependencies

All required Python packages are listed in requirements.txt:

langchain-community>=0.0.10
psutil>=5.9.0
numpy>=1.24.0
torch>=2.0.0  # Optional, for GPU support

Installation 💿

Automated Installation

# Make setup script executable
chmod +x setup.sh

# Run setup script
sudo ./setup.sh

Manual Installation

# Create virtual environment
python3 -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

Scripts Overview 📜

1. setup.sh

Main installation and configuration script:

• Updates system packages
• Installs dependencies
• Sets up Python environment
• Configures Ollama
• Runs initial system analysis

2. system_check.sh

Analyzes system capabilities:

• CPU information and extensions
• Memory capacity and speed
• GPU detection and specifications
• Storage availability
• Power management status
• Temperature sensors

3. select_model.sh

Recommends optimal model based on hardware:

• Analyzes system resources
• Suggests appropriate model size
• Provides configuration recommendations
• Adapts to laptop/desktop environments

4. optimize_system.sh

System optimization script:

• CPU governor management
• Memory optimization
• Power management (laptop-specific)
• Process priority optimization
• Thermal management

5. monitor.sh

Real-time system monitoring:

• CPU usage and temperature
• Memory utilization
• GPU statistics (if available)
• Process status
• Power consumption (laptops)

6. llm_benchmark.py

Performance testing suite:

• Response time measurement
• Memory usage tracking
• Token generation speed
• Temperature monitoring
• Resource utilization analysis

Configuration ⚙️

Model Selection

Available models with hardware requirements:

Model	RAM Required	Best For	Command
DeepSeek Coder 6.7B	6-8GB	Code completion, lightweight usage	ollama pull deepseek-coder:6.7b
DeepSeek LLM 7B	8GB	General purpose, efficient	ollama pull deepseek-llm:7b
DeepSeek R1 8B	12GB	Balanced performance	ollama pull deepseek-r1:8b
DeepSeek R1 14B	16GB	Enhanced capabilities	ollama pull deepseek-r1:14b
DeepSeek Coder V2 7B	8GB	Enhanced code completion	ollama pull deepseek-coder-v2:7b

Modelfile Configuration

# Dynamic model selection based on system memory
FROM {{ if gt .Memory 32 }}deepseek-r1:14b{{ else if gt .Memory 16 }}deepseek-r1:8b{{ else }}deepseek-coder:6.7b{{ end }}

# System-specific parameters
PARAMETER num_ctx {{ if gt .Memory 32 }}4096{{ else if gt .Memory 16 }}2048{{ else }}1024{{ end }}
PARAMETER num_predict {{ if gt .Memory 32 }}512{{ else if gt .Memory 16 }}256{{ else }}150{{ end }}

# Quality parameters
PARAMETER temperature 0.7
PARAMETER top_k 40
PARAMETER top_p 0.9
PARAMETER repeat_penalty 1.1

Usage 🎯

1. System Analysis

./system_check.sh > system_info.txt

2. Model Selection

./select_model.sh > model_selection.txt

3. System Optimization

sudo ./optimize_system.sh

4. Monitoring

./monitor.sh

5. Benchmark Testing

python llm_benchmark.py

Optimization 🚀

Desktop Systems

• Full CPU utilization
• Maximum performance governor
• High priority process scheduling
• GPU acceleration when available

Laptop Systems

• Dynamic CPU governor
• Thermal-aware processing
• Power-efficient thread allocation
• Battery life optimization

Monitoring 📊

Real-time monitoring includes:

• CPU usage and frequency
• Memory utilization
• GPU statistics
• Temperature tracking
• Power consumption
• Process statistics

Troubleshooting 🔧

Common Issues

1. Out of Memory

# Add swap space
sudo ./optimize_system.sh

2. High Temperature

# Monitor temperature
./monitor.sh

3. Poor Performance

# Check system status
./system_check.sh
# Adjust model selection
./select_model.sh

Contributing 🤝

1. Fork the repository
2. Create your feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License 📄

MIT License - See LICENSE file for details