- 🔭 I am currently working on a new gem "sqa" to do stock qualitative analysis on portfolios. I don't know which is easier: predicting the next number to appear on the roulette wheel or forecasting individual stock price movements 2 weeks (or even one day) in advance. I've been thinking about a probability cone over time much like the national hurricane center forcasts the track of storms.
- 🔭 I am also working on a context management system for large language models which is both temporal as well as within a semantic distance which mimics the human brain's hierarchical-temporal memory structure like short-term, working0 and long-term memory based upon either gross recollections or detailed recollections.
