Learn how to create a more advanced RAG pipeline in Kernel Memory using a combined Keyword + Vector Search followed by a ReRanking stage, actually implemented by Cohere really good ReRanking API.
The extension that made this possible is here: github.com/alkampfergit/Kerne...
Here is a previous video with some theory on the subject: • Kernel Memory - Nlp, P...
This is the repo containing the code: github.com/alkampfergit/Kerne...
▬ Contents of this video ▬▬▬▬▬▬▬▬▬▬
00:00 - Implementing Re-ranking for a Query
01:26 - Using Custom Kernel Memory Elasticsearch Extension
02:00 - Some live example on my security books library index
04:55 - Comparing BM25 and Vector Search Results
06:07 - Understanding the Role of Re-ranking in Search Results
10:50 - Configuring Kernel Memory
12:06 - Configuring the User Search Pipeline
13:34 - How the code is structured.
29 июл 2024