LangChainRB

Build Ruby-based Retrieval-Augmented Generation applications with LangChainRB and Zilliz Cloud or Milvus Vector Database

Use this integration for Free

LangChainRB and Milvus
LangChainRB is an orchestration layer for building Ruby-based LLM applications and is helpful to string multiple types of systems together, like vector databases (Milvus and Zilliz Cloud), and LLMs. With this solution, you can build semantic search capabilities that are useful for Q&A over documents, chatbots, and agents.

Key features include:
- Prompt Management — create, load and save prompt templates
- Context length validations — validate the context window length to avoid hitting LLM errors
- Data chunking — splitting the data into chunks that make semantic sense from various documents before creating embeddings that get stored in a vector database like Zilliz Cloud.
- Conversation memory — persisting a chat with an LLM to memory
- Provides support for popular LLMs like Anthropic, Cohere, Google Palm, Hugging Face, Local Llama, and OpenAI.
Benefits
- Common domain specific language and APIs
- Interoperability & reduced vendor lock in, pick your favorite LLM
- Best practices & rich feature set

How to execute a vector search with Milvus

llm = Langchain::LLM::GooglePalm.new(api_key: “...”)

client + Langchain::Vectorsearch::Milvus.new(url:, index_name:, llm: llm)
client.add_texts texts: [...], ids: [...]

my_pdf = Langchain.root.join(“/Documents/file.pdf”)
my_text = Langchain.root.join(“/Documents/file.txt”)
my_docx = Langchain.root.join(“/Documents/file.docx”)

client.add_data(paths: [my_pdf, my_txt, my_docx])

Some notes about the above example

This example is using Google Palm for the LLM, you can instantiate the LLM by passing it your API key
Create a client using the Milvus ruby gem, pass it the URL where Milvus is running, name the index and pass the index
Send vectors one by one to Milvus or bulk upsert

ActiveRecord example

Class Product < ActiveRecord::Base
   vectorsearch provider: Langchain::Vectorsearch::Milvus.new(url:, index_name:, llm:)

   After_save :upsert_to_vectorsearch
end

Vector search example

# Retrieve similar documents based on the query string passed in 
client.similarity_search(
     Query:,
     k:     # number of results to be retrieved
)

# Q&A-style querying based on the question passed in
client.ask(
      question:
)

Learn about LangChainRB and Milvus Integration

LangChainRB

LangChainRB and Milvus

Key features include:

Benefits

How to execute a vector search with Milvus

Learn about LangChainRB and Milvus Integration

Related Resources

Build AI Apps with Retrieval Augmented Generation (RAG)

Evaluating Your Embedding Model

A Guide to Methodologies, Metrics, and Evaluation Tools for Enhanced Reliability