- Contextual Retrieval In AI systems
- Hybrid search in vespa series
- HopRAG
- not able to implement it completely but it is in my backlog
- Lumber chunker
- Effective way of dynamic chunking
- I have optimized this approach to make production ready
- Cohere Reranking
- later went with Qwen Reranking
- LLM Embeddings Explained: A visual Guide
- SPLADE sparse vectors
This is one of the important concept I have learned while building the RAG pipeline. i,e., SAGA orchestration. A beautiful concepts.
- Retrieval Augmented Generation (RAG) Course
- This is the best course to learn about RAG for production case.
- 1st pass has been done, notes making under progress.
- On going work
- Building the deep research kind of agents to do QnA on documents. still in exploration phase.
- pageindex
- vectorless reasoning based RAG
- Very awesome idea, human like search, their demo page has really long documents.
- 37 insights from Weaviate Engineer on retrieval
- very good read on RAG's blind spot
- White-Paper: Applying embedding based retrieval to Airbnb search
- I have completed the first pass of the paper. I found a very interesting approach where they used IVF and Euclidean distance.
- This removed my bias toward always using cosine similarity. I still need to understand the exact usage, but it helped reduce my confirmation bias.
- Production ready RAG pipeline substack
- need to understand approach