Supercharge eCommerce Search: OpenAI's CLIP, BM25, and Python

Thanks for sharing this, James, we've already started implementing this idea in our app, after watching your video. Ofc we're using Pinecone.

Good job ! :)

This channel is shockingly good for its subscriber count. Lucky I found you. Thanks!

Check the new blip model which is basically chatgpt + clip I think. Also waiting for the arxiv project on langchain

Amazing content as always. I was wondering, is it recommended to use embeddings such as the ones form Openai or cohere instead of BM25?

Well explained, interesting!

Keep up the fantastic content mate.

A demo of what we are about to learn in the beginning of the video would greatly help an infant such as myself in this field.

This demo is fascinating. I would love to learn what technology to add to extend the demo, to maintain context between queries.

Great stuff!

HELP

Hi thanks for sharing the video it is really useful. For this type of usage, other the Pinecone are there any other vector DB that run offline on local machine?

This video is great! Instead of running on Colab, could you make a video that shows an up and down connection from an html front end to the Pinecone database, specifically uploading a PDF, vectoring it, querying, and displaying the results back through html? I also emailed you for some consulting work on a project. Thanks for the videos!

very nice, the sparse and dense vector mix can apply to many sceanrios.

I'm using s1 pod and trying to create an hybrid index with 10k vectors.
Will there any pricing difference between using a dense vector index alone and using a dense+sparse vector index from pinecone side?

Is there a reason why you didn't use CLIP to generate both image and text embeddings?

Hello James, great content. I have 1 query. How do we handle the query "show me blue jeans under $50", this "under $50" value while building a search engine. If you can guide me, would much appreciate it, thank you.

Discord link?

Hi! Thanks for a great notebook and walkthrough! Question: why do we fit BM25 only to `productDisplayName` field in `bm25.fit(metadata['productDisplayName'])` and not to all concatenated metadata fields (elements of meta_batch) which we use to actually encode documents? Wouldn't we miss some of the keywords present in other columns but missing in `productDisplayName`?

I thought the whole point of TF-IDF was to see first which unique keywords there are and index them. So, if we fit BM25 only to `productDisplayName` won't we basically ignore all other keywords that are in metdata but missing in `productDisplayName`? Thanks!

nice explanaysh bro