Using the API

Install the API client library

typescript

npm install dewy-ts

Python

pip install dewy-client dewy-langchain

Connect to an instance of Dewy

typescript

import { Dewy } from 'dewy-ts';
const dewy = new Dewy({endpoint: “http://localhost:3000”})

Python

from dewy_client import Client

client = Client(base_url="http://localhost:3000")

Add documents

typescript

await dewy.kb.addDocument({
  collection: "main",
  url: 'https://arxiv.org/abs/2005.11401',
})

Python

from dewy_client import Client
from dewy_client.api.kb import add_document
from dewy_client.models import AddDocumentRequest
add_document.sync(client=client, body=AddDocumentRequest(
    url = "https://arxiv.org/pdf/2305.14283.pdf",
    collection_id=collection_id,
))

Retrieve document chunks for LLM prompting

typescript

const context = await dewy.kb.retrieveChunks({
  collection: "main",
  query: "tell me about RAG", 
  n: 10,
});

// Minimal prompt example
const prompt = [
  {
    role: 'system',
    content: `You are a helpful assistant.
    You will take into account any CONTEXT BLOCK that is provided in a conversation.
    START CONTEXT BLOCK
    ${context.results.map((c: any) => c.chunk.text).join("\n")}
    END OF CONTEXT BLOCK
    `,
  },
]

// Using OpenAI to generate responses
const response = await openai.chat.completions.create({
  model: 'gpt-3.5-turbo',
  stream: true,
  messages: [...prompt, [{role: 'user': content: 'Tell me about RAG'}]]
})

Python

from langchain_core.runnables import RunnableParallel, RunnablePassthrough
from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import ChatPromptTemplate

from dewy_langchain import DewyRetriever

retriever = DewyRetriever.for_collection("main", base_url="http://localhost:8000")
prompt = ChatPromptTemplate.from_messages(
    [
        (
            "system",
            """
            You're a helpful AI assistant. Given a user question and some retrieved content, answer the user question.
            If none of the articles answer the question, just say you don't know.

            Here is the retrieved content:
            {context}
            """,
        ),
        ("human", "{question}"),
    ]
)

def format_chunks(chunks):
    return "\n\n".join([d.page_content for d in chunks])

chain = (
    { "context": retriever | format_chunks, "question": RunnablePassthrough() }
    | prompt
    | llm
    | StrOutputParser()
)