Y
Hacker News
new
|
ask
|
show
|
jobs
by
matchagaucho
266 days ago
As context window sizes increase and token prices go down, it makes more sense to inject dynamic memories into context (and use RAG/vector stores for knowledge retrieval).