vectorflow
vectorflow copied to clipboard
dgarnitz
VectorFlow is a high volume vector embedding pipeline that ingests raw data, transforms it into vectors and writes it to a vector DB of your choice.
VectorFlow should support Cohere Embeddings. There are multiple embeddings models. We should support the most common and newest ones
VectorFlow has many logs spread of over different containers. We need these logs to be aggregated into a searchable form. One option could be to use Kibana with Elastic Search....
Vectorflow should be able to ingest raw data from Salesforce. Some open questions to explore prior to implementation: - can this we done through the existing API or does it...
Right now most vector DBs support uploading duplicate vectors as long as the ID is unique. It might be beneficial for vectorflow to 1) prevent the uploading of duplicate vectors...
*Highly recommended to do this [issue ticket first](https://github.com/dgarnitz/vectorflow/issues/104)* ## What To Build - An endpoint add to the api that allows the user to perform RAG with a given text-only...
## What To Build - An endpoint add to the `api` that allows the user to pass in a text-only search term, embed it with a supported model (right now...
When hitting the `/embed` endpoint with incorrect parameters a 500 internal server error is thrown not the error from within the application. Examples: - No pinecone credentials / incorrect pinecone...
Right now we are running Flask with Gunicorn using PM2. There are a few problems with this: - Flask is not fast - PM2 is designed for node JS, not...