mirror of
https://github.com/meta-llama/llama-stack.git
synced 2025-12-22 18:46:16 +00:00
5e9c394500
Implement OpenGauss vector database integration for Llama Stack with the following features: - Add OpenGaussVectorIOAdapter for vector storage and retrieval - Support native vector similarity search operations - Implement connection and query management with psycopg2 - Provide configuration template for easy setup - Add comprehensive unit tests The implementation allows Llama Stack users to leverage OpenGauss as an enterprise-grade vector database for RAG applications. Users can configure their environment through a simple YAML configuration and environment variables.
27 lines
810 B
YAML
27 lines
810 B
YAML
version: 2
|
|
distribution_spec:
|
|
description: "Custom configuration using Opengauss for vector storage and metadata"
|
|
providers:
|
|
inference:
|
|
- remote::together-openai-compat
|
|
vector_io:
|
|
- remote::opengauss
|
|
agents:
|
|
- inline::meta-reference
|
|
tool_runtime:
|
|
- inline::rag-runtime
|
|
files:
|
|
- inline::localfs
|
|
telemetry:
|
|
- inline::meta-reference
|
|
safety:
|
|
- inline::llama-guard
|
|
|
|
image_type: venv
|
|
additional_pip_packages:
|
|
- psycopg2-binary>=2.9.3
|
|
- pgvector>=0.2.0
|
|
- asyncpg>=0.27.0
|
|
# rm -rf ~/.llama/distributions/opengauss-demo
|
|
# uv run --with llama-stack llama stack build --template opengauss-demo --image-type venv
|
|
# uv run --env-file .env --with llama-stack llama stack run /home/gt/.llama/distributions/opengauss-demo/opengauss-demo-run.yaml
|