elizaos-plugin-local-ai 2.0.0a4

elizaOS Local AI Plugin - Local LLM inference using llama.cpp for text, embeddings, vision, and audio

elizaOS Local AI Plugin (Python)

Python implementation of the Local AI plugin for elizaOS, providing local LLM inference using llama.cpp.

Installation

pip install elizaos-plugin-local-ai

Requirements

• Python 3.11+
• llama-cpp-python
• Model files (GGUF format)

Usage

from elizaos_plugin_local_ai import LocalAIPlugin, LocalAIConfig, TextGenerationParams

# Create configuration
config = LocalAIConfig(
    models_dir="/path/to/models",
    small_model="DeepHermes-3-Llama-3-3B-Preview-q4.gguf",
)

# Initialize plugin
plugin = LocalAIPlugin(config)

# Generate text
result = plugin.generate_text(TextGenerationParams(
    prompt="Tell me a joke",
    max_tokens=100,
))

print(result.text)

Configuration

Environment variables:

• MODELS_DIR: Directory containing model files
• CACHE_DIR: Directory for caching
• LOCAL_SMALL_MODEL: Filename of small model
• LOCAL_LARGE_MODEL: Filename of large model
• LOCAL_EMBEDDING_MODEL: Filename of embedding model
• LOCAL_EMBEDDING_DIMENSIONS: Embedding vector dimensions (default: 384)

License

MIT