Who wants in?

Software development tattoo candidates from the #Pivotal #agile #design guide.

#RAG practitioners, what's your approach to semantic chunking? I'm naively considering a method that iterates through sequential pairs of sentence #embeddings to find outlier peaks in semantic distance—like this screenshot from Greg Kamradt—and splitting at the peaks, but my intuition is that single sentences often lack adequate context, eg pronoun referents, for an #LLM to perform accurate #vectorization.

Awarded a #LlamaIndex hat for being annoying at #RAGathon.