Class ChatMistralAI<CallOptions>

Integration with a chat model.

Type Parameters

  • CallOptions extends MistralAICallOptions = MistralAICallOptions

Hierarchy (view full)

Implements

Constructors

constructor

Properties

apiKey maxTokens modelName safeMode safePrompt streaming temperature topP endpoint? randomSeed?

Methods

completionWithRetry invocationParams withStructuredOutput

Constructors

constructor

Properties

apiKey

apiKey: string

The API key to use.

Default

{process.env.MISTRAL_API_KEY}

maxTokens

maxTokens: number

The maximum number of tokens to generate in the completion. The token count of your prompt plus max_tokens cannot exceed the model's context length.

modelName

modelName: string = "mistral-small-latest"

The name of the model to use.

Default

{"mistral-small-latest"}

safeMode

safeMode: boolean = false

⚠️ Deprecated ⚠️

use safePrompt instead

This feature is deprecated and will be removed in the future.

It is not recommended for use.

safePrompt

safePrompt: boolean = false

Whether to inject a safety prompt before all conversations.

Default

{false}

streaming

streaming: boolean = false

Whether or not to stream the response.

Default

{false}

temperature

temperature: number = 0.7

What sampling temperature to use, between 0.0 and 2.0. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.

Default

{0.7}

topP

topP: number = 1

Nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. Should be between 0 and 1.

Default

{1}

Optional endpoint

endpoint?: string

Override the default endpoint.

Optional randomSeed

randomSeed?: number

The seed to use for random sampling. If set, different calls will generate deterministic results.

Methods

completionWithRetry

  • completionWithRetry(input, streaming): Promise<AsyncGenerator<ChatCompletionResponseChunk, any, unknown>>

  • Calls the MistralAI API with retry logic in case of failures.

    Parameters

    • input: MistralAIChatCompletionOptions

      The input to send to the MistralAI API.

    • streaming: true

    Returns Promise<AsyncGenerator<ChatCompletionResponseChunk, any, unknown>>

    The response from the MistralAI API.

  • completionWithRetry(input, streaming): Promise<ChatCompletionResponse>

  • Parameters

    • input: MistralAIChatCompletionOptions
    • streaming: false

    Returns Promise<ChatCompletionResponse>

invocationParams

  • invocationParams(options?): Omit<MistralAIChatCompletionOptions, "messages">

  • Get the parameters used to invoke the model

    Parameters

    • Optional options: unknown

    Returns Omit<MistralAIChatCompletionOptions, "messages">

withStructuredOutput

  • withStructuredOutput<RunOutput>(outputSchema, config?): Runnable<BaseLanguageModelInput, RunOutput>

  • Type Parameters

    • RunOutput extends Record<string, any> = Record<string, any>

    Parameters

    • outputSchema: any
    • Optional config: any

    Returns Runnable<BaseLanguageModelInput, RunOutput>

  • withStructuredOutput<RunOutput>(outputSchema, config?): Runnable<BaseLanguageModelInput, {
        parsed: RunOutput;
        raw: BaseMessage;
    }>

  • Type Parameters

    • RunOutput extends Record<string, any> = Record<string, any>

    Parameters

    • outputSchema: any
    • Optional config: any

    Returns Runnable<BaseLanguageModelInput, {
        parsed: RunOutput;
        raw: BaseMessage;
    }>

Settings

Member Visibility

Theme

On This Page

Generated using TypeDoc