Class ConversationSummaryBufferMemory

Class that extends BaseConversationSummaryMemory and implements ConversationSummaryBufferMemoryInput. It manages the conversation history in a LangChain application by maintaining a buffer of chat messages and providing methods to load, save, prune, and clear the memory.

Example

// Initialize the memory with a specific model and token limit
const memory = new ConversationSummaryBufferMemory({
  llm: new ChatOpenAI({ modelName: "gpt-3.5-turbo-instruct", temperature: 0 }),
  maxTokenLimit: 10,
});

// Save conversation context to memory
await memory.saveContext({ input: "hi" }, { output: "whats up" });
await memory.saveContext({ input: "not much you" }, { output: "not much" });

// Load the conversation history from memory
const history = await memory.loadMemoryVariables({});
console.log({ history });

// Create a chat prompt using the conversation history
const chatPrompt = ChatPromptTemplate.fromMessages([
  SystemMessagePromptTemplate.fromTemplate(
    "The following is a friendly conversation between a human and an AI. The AI is talkative and provides lots of specific details from its context. If the AI does not know the answer to a question, it truthfully says it does not know.",
  ),
  new MessagesPlaceholder("history"),
  HumanMessagePromptTemplate.fromTemplate("{input}"),
]);

// Initialize the conversation chain with the model, memory, and prompt
const chain = new ConversationChain({
  llm: new ChatOpenAI({ temperature: 0.9, verbose: true }),
  memory: memory,
  prompt: chatPrompt,
});

Hierarchy

Implements

Constructors

constructor

Properties

aiPrefix humanPrefix llm maxTokenLimit memoryKey movingSummaryBuffer prompt summaryChatMessageClass

Accessors

memoryKeys

Methods

clear loadMemoryVariables predictNewSummary prune saveContext

Constructors

constructor

Properties

aiPrefix

aiPrefix: string = "AI"

humanPrefix

humanPrefix: string = "Human"

llm

llm: BaseLanguageModelInterface

maxTokenLimit

maxTokenLimit: number = 2000

memoryKey

memoryKey: string = "history"

movingSummaryBuffer

movingSummaryBuffer: string = ""

prompt

prompt: BasePromptTemplate = SUMMARY_PROMPT

summaryChatMessageClass

summaryChatMessageClass: (new (content) => BaseMessage) = SystemMessage

Type declaration

    • new (content): BaseMessage

    • Parameters

      • content: string

      Returns BaseMessage

Accessors

memoryKeys

Methods

clear

  • Method that clears the memory and resets the movingSummaryBuffer.

    Returns Promise<void>

    Promise that resolves when the memory is cleared.

loadMemoryVariables

  • Method that loads the chat messages from the memory and returns them as a string or as a list of messages, depending on the returnMessages property.

    Parameters

    • Optional _: any

      InputValues object, not used in this method.

    Returns Promise<MemoryVariables>

    Promise that resolves with MemoryVariables object containing the loaded chat messages.

predictNewSummary

  • Predicts a new summary for the conversation given the existing messages and summary.

    Parameters

    • messages: BaseMessage[]

      Existing messages in the conversation.

    • existingSummary: string

      Current summary of the conversation.

    Returns Promise<string>

    A promise that resolves to a new summary string.

prune

  • Method that prunes the memory if the total number of tokens in the buffer exceeds the maxTokenLimit. It removes messages from the beginning of the buffer until the total number of tokens is within the limit.

    Returns Promise<void>

    Promise that resolves when the memory is pruned.

saveContext

  • Method that saves the context of the conversation, including the input and output values, and prunes the memory if it exceeds the maximum token limit.

    Parameters

    • inputValues: InputValues

      InputValues object containing the input values of the conversation.

    • outputValues: OutputValues

      OutputValues object containing the output values of the conversation.

    Returns Promise<void>

    Promise that resolves when the context is saved and the memory is pruned.

Settings

Member Visibility

Theme

On This Page

Generated using TypeDoc