Diary 2024-10-08
I have a suggestion to improve TTTC. After classifying the input with LLM, if we concatenate the embedding of the classification with the embedding of the original sentence and UMAP, we can get a nice diagram with separated clusters.
Processing in abstract high-dimensional space and visualizing it in two dimensions are two different requirements
Do you really want to be two-dimensional?
Should observe Turbo prompts.
To prepare for the possibility of radical change, practice in advance assuming that radical change has already occurred.
---
This page is auto-translated from /nishio/日記2024-10-08 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.