Memory is not recall: two 90-day runs to understand "remembering"
I published a memory benchmark scoring 8/12 over 90 days (10M tokens), thinking I had a recall problem. I actually had three more, hidden behind it.
15 min read
Builds Miraviel, an interactive story app, mostly at night. Writes here about what long-run testing reveals on AI memory, companions and roleplay, and about where all of this is heading.
I published a memory benchmark scoring 8/12 over 90 days (10M tokens), thinking I had a recall problem. I actually had three more, hidden behind it.
15 min read