#llm-wiki
3 items tagged llm-wiki.
Three architectures of agent memory — and why Trail picked Compile
Karpathy, Tan, and Liu all start from the same diagnosis — your agent is a retriever, not a thinker. They reach three different architectures: retrieve (RAG), compile (LLM Wiki), and act (Fat Skills / GBrain). Trail picked Compile, deliberately. This is what we accept and what we believe is converging.
Scaling Trail from 200 to 100,000 Neurons: An Engineering Note
A compile-time knowledge engine has three workloads that fight for compute — ingest, lint, and the curation queue. Each one breaks at a different corpus size, for different reasons, and requires a different fix. This is the long-form companion to Work That Fits in a Night: the full accounting of where Trail's bottlenecks are, when they hit, and what a deployment looks like at 200, 8,000, 25,000, and 100,000 Neurons.
Work That Fits in a Night: How Trail Stops Caring About Corpus Size
A compile-time knowledge system has to integrate new sources against everything it already knows. That integration cost grows with the corpus — until, one Tuesday, it stops finishing before the next night begins. Here is how Trail makes that cost bounded without going blind to the long tail.