AssistantMessage class
A message sent by the assistant.
- Extends
Constructors
| Assistant |
Creates a new 'AssistantMessage' instance. |
Inherited Properties
| required | If true the section is mandatory otherwise it can be safely dropped. |
| role | |
| separator | |
| template | |
| text |
|
| tokens | The requested token budget for this section.
|
Inherited Methods
| get |
Returns the content of a message as a string. |
| render |
Renders the prompt section as a string of text. |
Constructor Details
AssistantMessage(string, number, string)
Creates a new 'AssistantMessage' instance.
new AssistantMessage(template: string, tokens?: number, assistantPrefix?: string)Parameters
- template
-
string
Template to use for this section.
- tokens
-
number
Optional. Sizing strategy for this section. Defaults to -1, auto.
- assistantPrefix
-
string
Optional. Prefix to use for assistant messages when rendering as text. Defaults to assistant: .
Inherited Property Details
required
If true the section is mandatory otherwise it can be safely dropped.
required: booleanProperty Value
boolean
Inherited From TemplateSection.required
role
separator
template
textPrefix
tokens
The requested token budget for this section.
- Values between 0.0 and 1.0 represent a percentage of the total budget and the section will be layed out proportionally to all other sections.
- Values greater than 1.0 represent the max number of tokens the section should be allowed to consume.
tokens: numberProperty Value
number
Inherited From TemplateSection.tokens
Inherited Method Details
getMessageText(Message<string>)
Returns the content of a message as a string.
static function getMessageText(message: Message<string>): stringParameters
- message
-
Message<string>
Message to get the text of.
Returns
string
The message content as a string.
Inherited From TemplateSection.getMessageText
renderAsText(TurnContext, Memory, PromptFunctions, Tokenizer, number)
Renders the prompt section as a string of text.
function renderAsText(context: TurnContext, memory: Memory, functions: PromptFunctions, tokenizer: Tokenizer, maxTokens: number): Promise<RenderedPromptSection<string>>Parameters
- context
-
TurnContext
Context for the current turn of conversation.
- memory
- Memory
Interface for accessing state variables.
- functions
- PromptFunctions
Functions for rendering prompts.
- tokenizer
- Tokenizer
Tokenizer to use for encoding/decoding text.
- maxTokens
-
number
Maximum number of tokens allowed for the rendered prompt.
Returns
Promise<RenderedPromptSection<string>>
The rendered prompt section.
Inherited From TemplateSection.renderAsText