Namespace for language model related functionality.
Variables
onDidChangeChatModels:Event<Void>
An event that is fired when the set of available chat models changes.
tools:ReadOnlyArray<LanguageModelToolInformation>
A list of all available tools that were registered by all extensions using {@link lm.registerTool}. They can be called
with {@link lm.invokeTool} with input that match their declared inputSchema.
Methods
invokeTool(name:String, options:LanguageModelToolInvocationOptions<{}>, ?token:Null<CancellationToken>):Thenable<LanguageModelToolResult>
Invoke a tool listed in {@link lm.tools} by name with the given input. The input will be validated against the schema declared by the tool
A tool can be invoked by a chat participant, in the context of handling a chat request, or globally by any extension in any custom flow.
In the former case, the caller shall pass the {@link LanguageModelToolInvocationOptions.toolInvocationToken toolInvocationToken}, which comes with the a {@link ChatRequest.toolInvocationToken chat request}. This makes sure the chat UI shows the tool invocation for the correct conversation.
A tool {@link LanguageModelToolResult result} is an array of {@link LanguageModelTextPart text-} and
{@link LanguageModelPromptTsxPart prompt-tsx}-parts. If the tool caller is using @vscode/prompt-tsx, it can
incorporate the response parts into its prompt using a ToolResult. If not, the parts can be passed along to the
{@link LanguageModelChat} via a user message with a {@link LanguageModelToolResultPart}.
If a chat participant wants to preserve tool results for requests across multiple turns, it can store tool results in the {@link ChatResult.metadata} returned from the handler and retrieve them on the next turn from {@link ChatResponseTurn.result}.
@link CancellationTokenSource} for how to create one.
Parameters:
name | The name of the tool to call. |
|---|---|
options | The options to use when invoking the tool. |
token | A cancellation token. See { |
Returns:
The result of the tool invocation.
registerTool<T>(name:String, tool:LanguageModelTool<T>):Disposable
Register a LanguageModelTool. The tool must also be registered in the package.json languageModelTools contribution
point. A registered tool is available in the {@link lm.tools} list for any extension to see. But in order for it to
be seen by a language model, it must be passed in the list of available tools in {@link LanguageModelChatRequestOptions.tools}.
@link Disposable} that unregisters the tool when disposed.
Returns:
A {
selectChatModels(?selector:Null<LanguageModelChatSelector>):Thenable<Array<LanguageModelChat>>
Select chat models by a {@link LanguageModelChatSelector selector}. This can yield multiple or no chat models and extensions must handle these cases, esp. when no chat model exists, gracefully.
const models = await vscode.lm.selectChatModels({ family: 'gpt-3.5-turbo' });
if (models.length > 0) {
const [first] = models;
const response = await first.sendRequest(...)
// ...
} else {
// NO chat models available
}A selector can be written to broadly match all models of a given vendor or family, or it can narrowly select one model by ID. Keep in mind that the available set of models will change over time, but also that prompts may perform differently in different models.
Note that extensions can hold on to the results returned by this function and use them later. However, when the {@link onDidChangeChatModels}-event is fired the list of chat models might have changed and extensions should re-query.
Parameters:
selector | A chat model selector. When omitted all chat models are returned. |
|---|
Returns:
An array of chat models, can be empty!