AI Applications
This category is available for both desktop and headless installations, though some options are specific to desktop environments. All AI applications require CPU AVX support and will not be available on systems lacking this capability.
Security Focus: The AI packages listed below utilize a locally running AI API, which enhances privacy and security by processing data directly on your PC. This approach minimizes reliance on external services, ensuring that your data remains secure and under your control. However, if preferred, you have the option to switch to the OpenAI API.
For detailed configuration options and optimal settings for each tool, please refer to their respective documentation pages linked below.
Desktop-Specific AI Tools
The following tools are only available when installing a desktop environment:
Ditana Assistant GUI application
Ditana Assistant (AGPL-3.0-or-later) provides AI-powered system interaction through:
- Natural language interface for common terminal commands
- File operation assistance
- System query handling
- Basic troubleshooting support
- Desktop application launching
- Internet search capabilities
- Optional Integration of Wolfram|Alpha to provide real-time data.
In its default configuration, the assistant runs locally and prioritizes privacy by requesting confirmation before executing any command. You can easily configure it to use OpenAI models, see documentation.
Example screenshot, illustrating the seamless integration of real-time data from the Wolfram|Alpha API, which is automatically accessed as needed:
Logseq
Logseq (AGPL-3.0-or-later) is a powerful knowledge management platform that offers:
- Networked thought organization
- Daily journaling capabilities
- Task and project management
- Bi-directional linking
- Graph visualization
- Markdown and Org-mode support
- Local-first architecture ensuring data privacy
- Extensible plugin system
This Ditana-specific package extends Logseq with AI capabilities through:
- Pre-configured GPT-3 OpenAI plugin to make it use the local KoboldCpp instance
- AI-assisted note-taking capabilities
- Content generation features
- Knowledge synthesis support
This integration maintains privacy by leveraging your local KoboldCpp installation while providing the power of AI-enhanced knowledge management.
Example after pressing Ctrl+G
on a Logseq note:
Stable Diffusion
Stable Diffusion (easydiffusion) provides local AI image generation:
- Text-to-image generation
- High-quality output
- Local processing
- Custom model support
- Various generation parameters
- Intuitive interface
- Progress visualization
Core AI Tools
Ditana Assistant Terminal Application
Ditana Assistant (AGPL-3.0-or-later) provides AI-powered system interaction through:
- Natural language interface for common terminal commands
- Terminal-based interaction methods
- File operation assistance
- System query handling
- Basic troubleshooting support
- Desktop application launching
- Internet search capabilities
In its default configuration, the assistant runs locally and prioritizes privacy by requesting confirmation before executing any command. You can easily configure it to use OpenAI models, see documentation.
Example session:
❯ ./ditana-assistant “What are the largest files in the current directory and below?”
find . -type f -exec du -h {} + | sort -rh | head -n 10
Execute above command? (y/n) y
364K ./packaging/arch/ditana-assistant-1.036-1-any.pkg.tar.zst 284K ./src/ditana-logo.png 284K ./packaging/arch/pkg/ditana-assistant/usr/share/ditana/assistant/ditana-logo.png 284K ./packaging/arch/ditana-logo.png 276K ./.git/objects/21/e3382f3ee049d91368ce79218cb4049bae6222 248K ./myenv/lib/python3.12/site-packages/__pycache__/bottle.cpython-312.pyc 248K ./myenv/bin/__pycache__/bottle.cpython-312.pyc 204K ./myenv/lib/python3.12/site-packages/pip/_vendor/rich/__pycache__/_emoji_codes.cpython-312.pyc 160K ./myenv/lib/python3.12/site-packages/pip/_vendor/pkg_resources/__pycache__/__init__.cpython-312.pyc 156K ./myenv/lib/python3.12/site-packages/pip/_vendor/idna/__pycache__/uts46data.cpython-312.pyc
Your Message (’enter’ to quit): Which one of these are safe to delete?
It is important to exercise caution when deleting files, especially system files or files related to applications. In the list provided, the file that appears to be safe to delete without causing any major issues is:
./myenv/lib/python3.12/site-packages/pip/_vendor/idna/__pycache__/uts46data.cpython-312.pyc
This file seems to be a Python cache file and deleting it should not have a significant impact on the functionality of the application. However, it is always recommended to make a backup of the file before deleting it, in case it is needed later on.
See documentation for more examples.
KoboldCpp
KoboldCpp (AGPL-3.0-only) offers private, local AI chat functionality:
- Browser-based interface
- Pre-configured Gemma language model
- Text generation capabilities
- Code completion support
- Local web server implementation
- Privacy-focused design
- Suitable for both desktop and headless systems
Ditana installs it preconfigured with the Gemma language model. A desktop shortcut is installed to access it via its browser interface. This package is also used by the Ditana Assistant via the KoboldCpp API.
Development Tools
CodeGPT
CodeGPT is a development tool that integrates AI assistance into your workflow:
- Automated git commit message generation
- AI-powered code review capabilities
- OpenAI API integration
- Command-line interface
- Multiple model support
- Context-aware suggestions
- Integration with popular git commands
The Ditana installer configures CodeGPT to use gpt-4o-mini instead of gpt-3.5-turbo. As of October 2024, gpt-4o-mini is a factor of 20 cheaper than gpt-3.5-turbo: $0.150 / 1M input tokens. According to https://platform.openai.com/docs/models/gpt-3-5-turbo:
“As of July 2024, gpt-4o-mini should be used in place of gpt-3.5-turbo, as it is cheaper, more capable, multimodal, and just as fast. gpt-3.5-turbo is still available for use in the API.”
Note: Requires an OpenAI API key stored in ~/.shell.d/openai.sh
High-Performance Computing Libraries
A comprehensive suite of libraries optimized for AI/ML workloads and numerical computations:
Basic Libraries
- blas-openblas: Basic Linear Algebra Subprograms for matrix operations
- clblast: OpenCL BLAS library for GPU acceleration
- ocl-icd, opencl-clhpp: OpenCL standard implementations
- openmpi: High-performance message passing library
NVIDIA-Specific Components
When NVIDIA GPU is detected:
- cuda: NVIDIA’s parallel computing platform
- cudnn: Deep Neural Network library
- opencl-nvidia: NVIDIA OpenCL implementation
- lib32-opencl-nvidia: 32-bit NVIDIA OpenCL support
- nccl: Multi-GPU/node collective communication
AMD-Specific Components
For AMD GPU systems:
- hipblas: ROCm BLAS library
- hip-runtime-amd: AMD’s compute interface
System Requirements
- CPU must support AVX instructions
- Adequate RAM (varies by tool)
- Sufficient storage space
Feedback
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.