Explore an advanced tree retrieval system inspired by the Memwalker paper in this 26-minute video tutorial. Learn how to implement a custom Python code that utilizes GPT-3.5 Turbo for summarization and GPT-4 for reasoning, navigation, and retrieval without relying on embeddings or VectorDBs. Follow along as the instructor demonstrates the construction of a JSON tree structure, runs the code line-by-line for tree construction and retrieval, and discusses memory considerations from the original paper. Gain insights into this innovative approach to information retrieval and its potential applications in artificial intelligence and large language models.
Memwalker Advanced Tree Retrieval Without VectorDB - Python Implementation
Overview
Syllabus
Introduction
overview of JSON tree structure
Demo of constructing the tree
echohive projects
Running the code line by line for tree construction
Running the code line by line for retrieval and GPT answer
Memory and other consideration for the original paper
Taught by
echohive
