Ziad's blog

Using ordinary LLMs such as ChatGPT or LLAMA for extracting relations between entities in text proves to be expensive and slow, but what if we used a Seq2Seq model, which are sort of a hybrid between GPTs and BERTs, the latter which is still used for relation extraction. Could we make it cheaper and faster?

Building a RAG system using Cluster Semantic Chunking as recommended by ChromaDB researches, while using PGVector as the vector database. Demo website available

Building a practical analytics-ready URL shortening app using Spring boot, React w/ Shadcn & Docker on a VPS