Overview

Application calculates cosine similarity between reference database and incoming embeddings.
It returns recommended items from reference database with their probabilities (cosine similarity score).

Application key points:

• Return recommended item from reference database using API
• Load previously calculated embeddings from Embeddings Calculator application to RAM
• Allows dynamically updates reference database in RAM

How to Run

1. Add AI Recommendations to your team from Ecosystem.

AI Recommendations

app

Recommends matching items from the catalog

2. Run app from the context menu of Images Project:

ℹ️ You can use Embeddings Calculator application to get Images Project in suitable format

How to Use

1. Select checkpoint with which you computed the embeddings for the selected project
2. Load selected embeddings

Demo Video

Results

Embeddings have been loaded to RAM and are ready to go.
The application will continue to run in server mode and wait for incoming requests.

requests examples