Skip to content

Latest commit

 

History

History
53 lines (46 loc) · 2.19 KB

Milvus.md

File metadata and controls

53 lines (46 loc) · 2.19 KB

Milvus

If you wish to utilize the Milvus vector database within agentUniverse, you need to follow these steps:

  1. Install the Milvus vector database You can consult the official Milvus installation documentation for detailed instructions on installing and using Milvus. As a recommendation, you can start the Milvus container in Docker by executing the following commands:
# Download the installation script
$ curl -sfL https://raw.githubusercontent.com/milvus-io/milvus/master/scripts/standalone_embed.sh

# Start the Docker container
$ bash standalone_embed.sh start

These commands will download the Milvus image and initiate a container, thereby offering database services on port 19530. For additional information and alternative installation methods, kindly refer to the official documentation.

  1. Install the Milvus Python SDK
pip install pymilvus

How to Configure Milvus Components

name: 'milvus_store'
description: 'a store based on milvus'
connection_args:
  host: '127.0.0.1'
  port: '19530'
search_args:
  metric_type: "L2"
  params:
    nprobe: 10
index_params:
  metric_type: "L2"
  index_type: "HNSW"
  params:
    M: 8
    efConstruction: 32
embedding_model: 'dashscope_embedding'
similarity_top_k: 100
metadata:
  type: 'STORE'
  module: 'agentuniverse.agent.action.knowledge.store.milvus_store'
  class: 'MilvusStore'
  • connection_args: Parameters required for connecting to the Milvus database, encompassing the host address (host) and port number (port).
  • search_args: Search parameters, defining the type of distance metric (metric_type) utilized during searches, along with related parameters such as nprobe.
  • index_params: Indexing parameters, specifying the index type (index_type), distance metric type (metric_type), and specific parameters pertinent to index construction, such as M and efConstruction.
  • embedding_model: The model employed to generate embedding vectors, specified here as dashscope_embedding.
  • similarity_top_k: The number of the most similar results returned in a similarity search.

Usage

Knowledge_Define_And_Use