JS/TS
JS/TS Quickstart
Connect to vectorview with 3 lines of code
Install
npm install vectorview
Use
Get your key
by logging in to the vectorview dashboard
import Vectorview from "vectorview";
const vv = new Vectorview(key)
vv.event(query, docsWithScore)
Example
import { FaissStore } from "langchain/vectorstores/faiss";
import { OpenAIEmbeddings } from "langchain/embeddings/openai";
import { TokenTextSplitter } from "langchain/text_splitter";
import Vectorview from "vectorview";
// Setup
const vv = new Vectorview(key);
// Do semantic search
const text = await fs.readFile('./text.txt', 'utf8');
const splitter = new TokenTextSplitter({
chunkSize: 40,
chunkOverlap: 0,
});
const docs = await splitter.createDocuments([text]);
const vectorStore = await FaissStore.fromDocuments(docs, new OpenAIEmbeddings());
const query = "What did the president say about Ketanji Brown Jackson";
const docsWithScore = await vectorStore.similaritySearchWithScore(query, 3);
// Log vv event
vv.event(query, docsWithScore)
Advanced Use
Query metadata
Assign custom metadata to an event()
by passing a dict
as the third argument.
vv.event(query, docsWithScore, {"foo": "bar"})
Document metadata
Documents in vectorview
are langchain Documents which has a metadata
field. Any metadata added will show up in your vectorview
dashboard.
docsWithScore[0][0].metadata = {"foo": "bar"}
vv.event(query, docsWithScore)
Custom document id
Vectorview assigns an id to each document used in an event()
. To assign a custom id, add an id
field in the Document’s metadata.
docsWithScore[0][0].metadata = {"id": new_id}
vv.event(query, docsWithScore)