https://github.com/kunalvirwal/clip-vectorizer

A containerized API for generating Vector embeddings for text and images using the OpenAI CLIP Model utilizing CUDA
https://github.com/kunalvirwal/clip-vectorizer

clip containerized-services fastapi openai python python3 vector-embeddings

Last synced: 3 months ago
JSON representation

A containerized API for generating Vector embeddings for text and images using the OpenAI CLIP Model utilizing CUDA

• Host: GitHub
• URL: https://github.com/kunalvirwal/clip-vectorizer
• Owner: kunalvirwal
• Created: 2024-08-19T21:13:56.000Z (9 months ago)
• Default Branch: main
• Last Pushed: 2024-09-02T16:35:54.000Z (9 months ago)
• Last Synced: 2025-01-11T18:45:37.513Z (5 months ago)
• Topics: clip, containerized-services, fastapi, openai, python, python3, vector-embeddings
• Language: Python
• Homepage:
• Size: 12.7 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# ***CLIP-Vectorizer***

This is openAI's CLIP model based API that creates text and image vector-embeddings to be stored and query a vector database.

## ***Steps To run on localhost using Docker***

- Make sure Docker is installed and running (and using WSL2 engine if in windows).

- Follow the steps given in [Nvidia docs](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html) to install nvidia drivers for your distribution (WSL for windows).

- If CUDA drivers are not present or GPU access is not provided to the container, then it will automatically default to computing on CPU.

- This command builds the image to be run inside a container  

  > `docker build -t vectorizer .`

- Run the program inside a container using

  > `docker run -it --gpus all -p 5000:8080 vectorizer`  

## ***API routes***

- `/vectors`   

  > Post route for sending text to be embedded in JSON format.   

  > Example Input JSON:  

  > {  

  >  "text" : "Your text here",  

  > }

- `/vectors_img`  

  > Post route for sending images to be embedded in form-data format.   

  > Example Input Form-data:  

  > Key : file | Value : (Your image file)