Expand description
Memory backends for agentrs.
Structs§
- Approximate
Tokenizer - Lightweight tokenizer approximation that avoids external dependencies.
- InMemory
Memory - Default in-process memory backend.
- InMemory
Vector Store - In-memory vector store with cosine similarity search.
- Simple
Embedder - Small deterministic embedder useful for tests and local demos.
- Sliding
Window Memory - Memory backend that keeps a fixed number of recent non-system messages.
- Token
Aware Memory - Memory backend that trims history to fit a token budget.
- Vector
Memory - Memory backend that combines recent history with semantic retrieval.
- Vector
Search Result - Search result returned by a vector store.
Traits§
- Embedder
- Computes embeddings for messages.
- Searchable
Memory - Shared extension trait for memory utilities.
- Tokenizer
- Counts approximate tokens for a string.
- Vector
Store - Persists and searches embedding vectors.