At this moment this sample app embeds one document only at a time. You can extend and adopt to your needs.

docker-compose -f docker-compose.yml up -d

• Navigate browser to http://localhost:8010/docs
• Create Tenant http://localhost:8010/docs#/default/create_tenant-v2 E.g name - lhc

curl -X 'POST' \
  'http://127.0.0.1:8010/api/v2/tenants' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "name": "lhc"
}'

• Create Database http://localhost:8010/docs#/default/create_database-v2 by entering Tenant name lhc

curl -X 'POST' \
  'http://127.0.0.1:8010/api/v2/tenants/lhc/databases' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "name": "lhc"
}'

To list databases

curl -X 'GET' 'http://localhost:8010/api/v2/tenants/lhc/databases' -H 'accept: application/json'

To list collections

curl -X 'GET' 'http://localhost:8010/api/v2/tenants/lhc/databases/lhc/collections' -H 'accept: application/json'

Instrutions here

https://github.com/LiveHelperChat/rag-embed?tab=readme-ov-file#introduction

E.g in the sample-application folder I just put my documentation file. Documentation file was generated using

docker run --rm -v $(pwd)/output:/app/output -v $(pwd)/cache:/app/cache remdex/crawler-to-md --url https://doc.livehelperchat.com --exclude "/docs/hooks"

More information at https://github.com/LiveHelperChat/crawler-to-md

php embed_documents.php httpsdoc.livehelperchat.com.md

After that in this folder you will see output.json file generated. Generation for 1.3 MB files takes about 30 seconds.

php store_embeddings.php

You should see something like

Embedding dimension: 384
Using existing collection: lhc with id bb308c00-4392-4f8b-ba81-6f722221fc64
Successfully stored 3459 embeddings to ChromaDB

Write down your collection id bb308c00-4392-4f8b-ba81-6f722221fc64

Last argument is your collection ID

php search_embeddings.php "Your search query" "Collection ID"

Flow should look like

• First query should go to embeding server and embedings retrieved
• Embedings should be send to Vector Database Chroma DB
• Embedings results merged into single text
• Embedings should be send to LLM for final answer