Uploaded September 2024 | Updated September 2026, 5 hours ago
RAG Applications are a popular way to incorporate AI into your use cases and make your data more accessible.
To optimize your RAG applications, consider RAG a framework where each portion of the R, A, and G pipeline can be improved, evaluated, and assessed!
In this workshop, we focus on different chunking strategies in the context of RAG Application.
We explore basic techniques like Character Splitting or Recursive Character Splitting and discuss some common challenges and considerations, such as the optimal selection of chunk size and overlap windows.
Together, we also explore semantic chunking techniques that dynamically adjust based on textual meaning and discuss the use of LLM-based chunking to automate chunk creation.
To add to all of this, we discuss Small2Big, a method that uses different chunks for retrieval versus generation, and demonstrate how leveraging metadata from chunks can refine search results in RAG systems.
If you want to discuss more topics like this with other community members, we love to invite you to our Community RAG Corner: weaviate.slack.com/archives/C07EJS6LQVA.
We are looking forward to dive into this topic with you.s
RAG Applications are a popular way to incorporate AI into your use cases and make your data more accessible.
To optimize your RAG applications, consider RAG a framework where each portion of the R, A, and G pipeline can be improved, evaluated, and assessed!
In this workshop, we focus on different chunking strategies in the context of RAG Application.
We explore basic techniques like Character Splitting or Recursive Character Splitting and discuss some common challenges and considerations, such as the optimal selection of chunk size and overlap windows.
Together, we also explore semantic chunking techniques that dynamically adjust based on textual meaning and discuss the use of LLM-based chunking to automate chunk creation.
To add to all of this, we discuss Small2Big, a method that uses different chunks for retrieval versus generation, and demonstrate how leveraging metadata from chunks can refine search results in RAG systems.
If you want to discuss more topics like this with other community members, we love to invite you to our Community RAG Corner: weaviate.slack.com/archives/C07EJS6LQVA.
We are looking forward to dive into this topic with you.s
