new·The score now tells you which way it movedA brain's exam only ever grows: its own material writes questions, and so does every question a real caller asked and did not get answered. The score is a percentage over that growing set, so a brain that learned more could post a smaller number — and this week three did. One of them answered two MORE questions than the week before and showed eighteen points less. Printed as a single percentage, that reads as decline to a reader and as punishment to anyone who contributes material.all news →
mozg.beta
Sign in

LangChain & LangGraph · all subjects

memory/semantic

2 notes, read out of this brain and free to use. Each one was extracted from a source and is re-checked against its exam.

Semantic memory management via profile

In semantic memory management using a profile approach, memories are stored as a single, continuously updated JSON document with key-value pairs representing user, organization, or agent information. When updating a profile, pass in the previous profile and ask the model to generate a new profile or a JSON patch to apply. This approach can become error-prone as the profile gets larger and may benefit from splitting into multiple documents or using strict decoding to ensure the memory schema remains valid.

Semantic memory management via collection

In semantic memory management using a document collection approach, memories are stored as a collection of documents that are continuously updated and extended over time. Each individual memory is more narrowly scoped and easier to generate, leading to higher recall downstream compared to profile approaches. However, this shifts complexity to memory updating, as the model must delete or update existing items in the list. The Store supports both semantic search and filtering by content for searching over document collections.

Give your agent this brain