Great post. One application you should add is Vespa. It's my impression they have been leading the way in Hybrid Search (predating the emergence of Vector Databases). They have a wealth of knowledge on their blog and on youTube - to go along with their open source application. See https://blog.vespa.ai/vespa-hybrid-billion-scale-vector-search/
Christopher
ps. Also enjoyed your presentation at Haystack this year =)
This was very refreshing to read - I am new in this area and I kept asking myself: wait - but why exactly do I need embeddings? You made a good job bringing some clarity to the subject. But I think you still underestimate the complexity of the task - because you can mix not just vector, keyword and relational database search - but also you can use the LLM itself to guide these searches by finding the relevant keywords, choosing sources etc, possibly in a recursive way (by asking what else is needed for a given task). There is also one more trick that I know about: you can query an ungrounded LLM - and then use the generated answer to find the relevant keywords, vectors or other indexes. I suspect that there are many others.
When the whole industry is down, it invents a new hype to rally money and keep the lie going
I've delved deep into GenAI, LLMs, LangChain, vector store etc... Most biz people don't want to believe any limits because the media has established the false anthropomorphized belief in AI capabilities
Hi Collin,
Great post. One application you should add is Vespa. It's my impression they have been leading the way in Hybrid Search (predating the emergence of Vector Databases). They have a wealth of knowledge on their blog and on youTube - to go along with their open source application. See https://blog.vespa.ai/vespa-hybrid-billion-scale-vector-search/
Christopher
ps. Also enjoyed your presentation at Haystack this year =)
This was very refreshing to read - I am new in this area and I kept asking myself: wait - but why exactly do I need embeddings? You made a good job bringing some clarity to the subject. But I think you still underestimate the complexity of the task - because you can mix not just vector, keyword and relational database search - but also you can use the LLM itself to guide these searches by finding the relevant keywords, choosing sources etc, possibly in a recursive way (by asking what else is needed for a given task). There is also one more trick that I know about: you can query an ungrounded LLM - and then use the generated answer to find the relevant keywords, vectors or other indexes. I suspect that there are many others.
A practical question - how do you call this practice?
- Retrieval Augmented Generation
- In Context Querying
- Grounding
- something else
I like grounding - it is the shortest.
Came to this post from: https://youtu.be/oqW19BvZcGg
Thank you for taking the time to write this!
When the whole industry is down, it invents a new hype to rally money and keep the lie going
I've delved deep into GenAI, LLMs, LangChain, vector store etc... Most biz people don't want to believe any limits because the media has established the false anthropomorphized belief in AI capabilities
Great post. I would add MongoDB to the Hybrid search list as it supports mixing these two forms of queries (thanks to Lucene)
Thank you for a great article that provides a much needed perspective while media is flooded with a biased view.