https://github.com/comhendrik/vectormatch
Dockerized application that embeds text in a pgvecto.rs database and retrieves data with a similarity search to generate a response with an llm from ollama.
https://github.com/comhendrik/vectormatch
docker docker-compose embeddings-similarity nlp ollama pgvecto-rs postgresql python vector vector-database
Last synced: 2 months ago
JSON representation
Dockerized application that embeds text in a pgvecto.rs database and retrieves data with a similarity search to generate a response with an llm from ollama.
- Host: GitHub
- URL: https://github.com/comhendrik/vectormatch
- Owner: comhendrik
- Created: 2024-09-12T11:40:29.000Z (9 months ago)
- Default Branch: main
- Last Pushed: 2024-12-01T09:32:15.000Z (7 months ago)
- Last Synced: 2025-04-07T14:47:38.250Z (2 months ago)
- Topics: docker, docker-compose, embeddings-similarity, nlp, ollama, pgvecto-rs, postgresql, python, vector, vector-database
- Language: Python
- Homepage:
- Size: 31.3 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Text Embedding and Search with PostgreSQL and Hugging Face in Docker
This project demonstrates a Python script that embeds text using a model from Hugging Face, stores the embeddings in PostgreSQL with the `pgvector` extension, and allows searching the database using regular text queries by comparing embeddings. After the data is retrieved an llm is used to generate a response with ollama. The Project is run with Docker Compose
## Features
- **Embeddings:** Use Hugging Face's transformers to embed input text.
- **PostgreSQL with pgvector:** Store embeddings in a PostgreSQL database using the `pgvector` extension to perform vector-based searches.
- **Search Functionality:** Retrieve database entries by comparing the input text's embedding to the stored embeddings.
- **Docker Support:** Run the whole application with Docker compose
- **Ollama:** Generate response based on local llm
## Prerequisites
Make sure you have the following installed:
- **Docker**
### Setup
Get the project directory
```
git clone https://github.com/comhendrik/vectorMatch.git
```
Start docker and go into the project directory and run the compose file
```
docker compose up
```
Wait for the script to be done, this can take a few minutes and then attach yourself to the vectorMatch container
```
docker attach vectormatch-vector-match-1
```