3 ms·
Yeah, memory works by injecting everything it has "remembered" as part of the system prompt at the beginning of each new chat session: https://simonwillison.net
by simonw 2y ago
Yeah, memory works by injecting everything it has "remembered" as part of the system prompt at the beginning of each new chat session: https://simonwillison.net/2024/Feb/14/memory-and-new-controls-for-chatgpt/ https://simonwillison.net/2024/Feb/14/memory-and-new-control...
- joquarky 2y agoI'm curious how it will avoid filling up the context window as more memories are added.
- refulgentis 2y agoCount the tokens, RAG em down to max count you're willing to pay for
- hackable_sand 2y agoVector compression. "Summarize memories at timestamps: [i, j, k, ...n]" etc. etc. until you have a nice scrapbook of memories. Then you gotta prune low-traffic, low-value vector spaces by whatever heuristic you (or an automated system) deems important.