Implementation:Langchain ai Langchain BaseMemory

Knowledge Sources	Langchain_ai_Langchain
Domains	Memory, Chain State Management
Last Updated	2026-02-11 00:00 GMT

Overview

BaseMemory is a deprecated abstract base class that defines the interface for memory components in LangChain Chains, enabling state persistence across chain executions.

Description

BaseMemory is an abstract base class located in the langchain_classic package (formerly langchain v0.0.x). It extends both Serializable and Python's ABC, providing the foundational interface for memory management in Chains. Memory allows storing information about past executions and injecting that context into future runs, which is particularly useful for conversational chains that need to maintain conversation history.

This class has been deprecated since v0.3.3 and is scheduled for removal in LangChain v1.0.0. Users are directed to the migration guide for updated memory patterns.

Usage

Import this class when building custom memory implementations for legacy LangChain Chains. New projects should follow the migration guide at the LangChain documentation for current memory patterns.

Code Reference

Source Location

• Repository: Langchain_ai_Langchain
• File: libs/langchain/langchain_classic/base_memory.py
• Lines: 1-117

Signature

@deprecated(since="0.3.3", removal="1.0.0", message=("..."))
class BaseMemory(Serializable, ABC):
    model_config = ConfigDict(arbitrary_types_allowed=True)

    @property
    @abstractmethod
    def memory_variables(self) -> list[str]: ...

    @abstractmethod
    def load_memory_variables(self, inputs: dict[str, Any]) -> dict[str, Any]: ...

    async def aload_memory_variables(self, inputs: dict[str, Any]) -> dict[str, Any]: ...

    @abstractmethod
    def save_context(self, inputs: dict[str, Any], outputs: dict[str, str]) -> None: ...

    async def asave_context(self, inputs: dict[str, Any], outputs: dict[str, str]) -> None: ...

    @abstractmethod
    def clear(self) -> None: ...

    async def aclear(self) -> None: ...

Import

from langchain_classic.base_memory import BaseMemory

I/O Contract

Inputs

Name	Type	Required	Description
inputs	dict[str, Any]	Yes	The inputs to the chain, used by load_memory_variables and save_context.
outputs	dict[str, str]	Yes	The outputs of the chain, used by save_context.

Outputs

Name	Type	Description
memory_variables	list[str]	The string keys that this memory class adds to chain inputs.
load_memory_variables return	dict[str, Any]	Key-value pairs of memory data given the text input to the chain.

Abstract Methods

Subclasses must implement the following:

Method	Return Type	Description
memory_variables (property)	list[str]	Returns the string keys this memory class will add to chain inputs.
load_memory_variables(inputs)	dict[str, Any]	Returns key-value pairs given the text input to the chain.
save_context(inputs, outputs)	None	Saves the context of a chain run to memory.
clear()	None	Clears memory contents.

Async variants (aload_memory_variables, asave_context, aclear) have default implementations that delegate to their synchronous counterparts via run_in_executor.

Usage Examples

Basic Usage

from langchain_classic.base_memory import BaseMemory
from typing import Any


class SimpleMemory(BaseMemory):
    memories: dict[str, Any] = dict()

    @property
    def memory_variables(self) -> list[str]:
        return list(self.memories.keys())

    def load_memory_variables(self, inputs: dict[str, Any]) -> dict[str, str]:
        return self.memories

    def save_context(
        self, inputs: dict[str, Any], outputs: dict[str, str]
    ) -> None:
        pass

    def clear(self) -> None:
        pass

Related Pages

• Heuristic:Langchain_ai_Langchain_Warning_Deprecated_Langchain_Classic
• Deprecated since v0.3.3; see Memory Migration Guide