Prompting LLM's with 1M words with Dr Mikhail Burtsev

Turn ChatGPT, Bard, or any LLM from an amnesiac goldfish into a memory mammoth!

Listen now:
https://youtu.be/4QDvNiyAikw

We interview Dr Mikhail Burtsev, Landau AI Fellow at the London Institute for Mathematical Sciences (LIMS) and author of the paper, "Scaling Transformer to 1M tokens and beyond with RMT" (Mikhail S. Burtsev, Aydar Bulatov, Yuri Kuratov) https://arxiv.org/abs/2304.11062

Think pasting in a whole book series to ChatGPT and asking questions, vs the 'half an article' which was possible up until now. This brings the potential to get insights from much larger bodies of knowledge/text without any kind of loss of fidelity from a knowledge-base lookup or chunking approach. Dr Burtsev suggests that the real potential of this invention will only be fully realised in more challenging applications than simple LLM chat.

Also available in audio format on:

• Spotify: https://open.spotify.com/episode/32dGZm0fIcmnD8Mfj9yTQp?si=7srSUg3ZTeiW8EB7le-Qtg
• Apple Podcasts: https://podcasts.apple.com/us/podcast/nz-good-ai-works/id1680691271?i=1000622067609
• Your favourite podcast platform