video1.0<iframe src="https://www.loom.com/embed/64b14628cc5a4ca598c26f2f4eaaefbb" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/64b14628cc5a4ca598c26f2f4eaaefbb-ef16a6ce648dab08.gif946.22In this video, I walk through the process of extracting questions from a research paper on the challenges of alignment in large language models. I initially faced some issues with chunking the text correctly, but after debugging, I was able to extract around 225 questions successfully. I emphasize the importance of clear instructions for the language model to avoid hallucinations. Moving forward, I plan to implement tool calling to enhance our system's capabilities. I encourage viewers to consider how we can further refine our approach when we reconvene after lunch.