Clarification Needed on Few-Shot Method Implementation with new syntax

[catallo]

Hello everyone,

I'm currently working on a project where I need to use the few-shot method with 5 user/assistant prompts to make GPT-3.5 Turbo output what I need. After adopting to the latest changes in the library I've noticed that my code is now substantially longer than before. I wonder if this increase in code length is necessary or if I might be doing something wrong.

This is the old code:

void requestGPTexplain(String prompt) {
  String completeResponse = "";

  OpenAI.apiKey = apiKey ?? exit(1);

  Stream<OpenAIStreamChatCompletionModel> chatStream =
      OpenAI.instance.chat.createStream(
    model: model,
    messages: [
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExSystem,
        role: OpenAIChatMessageRole.system,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExUser1,
        role: OpenAIChatMessageRole.user,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExAssistant1,
        role: OpenAIChatMessageRole.assistant,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExUser2,
        role: OpenAIChatMessageRole.user,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExAssistant2,
        role: OpenAIChatMessageRole.assistant,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExUser3,
        role: OpenAIChatMessageRole.user,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExAssistant3,
        role: OpenAIChatMessageRole.assistant,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExUser4,
        role: OpenAIChatMessageRole.user,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExAssistant4,
        role: OpenAIChatMessageRole.assistant,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExUser5,
        role: OpenAIChatMessageRole.user,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        content: promptExAssistant5,
        role: OpenAIChatMessageRole.assistant,
      ),
      OpenAIChatCompletionChoiceMessageModel(
        //content: jsonEncode("$promptExUser$prompt"),
        content: promptExUser + prompt,
        role: OpenAIChatMessageRole.user,
      )
    ],
    temperature: temp,
  );

  chatStream.listen(
      (streamChatCompletion) {
        String? content = streamChatCompletion.choices.first.delta.content;

        stdout.write(content);
[ ... ]

Now with version 4.1.4 does it have to look like this?

void requestGPTexplain(String prompt) {
  final systemRole = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExSystemRole,
      ),
    ],
    role: OpenAIChatMessageRole.system,
  );

  final user1 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExUser1,
      ),
    ],
    role: OpenAIChatMessageRole.user,
  );

  final assist1 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExAssistant1,
      ),
    ],
    role: OpenAIChatMessageRole.assistant,
  );

  final user2 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExUser2,
      ),
    ],
    role: OpenAIChatMessageRole.user,
  );

  final assist2 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExAssistant2,
      ),
    ],
    role: OpenAIChatMessageRole.assistant,
  );

  final user3 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExUser3,
      ),
    ],
    role: OpenAIChatMessageRole.user,
  );

  final assist3 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExAssistant3,
      ),
    ],
    role: OpenAIChatMessageRole.assistant,
  );

  final user4 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExUser4,
      ),
    ],
    role: OpenAIChatMessageRole.user,
  );

  final assist4 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExAssistant4,
      ),
    ],
    role: OpenAIChatMessageRole.assistant,
  );

  final user5 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExUser5,
      ),
    ],
    role: OpenAIChatMessageRole.user,
  );

  final assist5 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExAssistant5,
      ),
    ],
    role: OpenAIChatMessageRole.assistant,
  );

  final user6 = OpenAIChatCompletionChoiceMessageModel(
    content: [
      OpenAIChatCompletionChoiceMessageContentItemModel.text(
        promptExUser + prompt,
      ),
    ],
    role: OpenAIChatMessageRole.user,
  );

  final chatStream = OpenAI.instance.chat.createStream(
    model: "gpt-3.5-turbo",
    messages: [
      systemRole,
      user1,
      assist1,
      user2,
      assist2,
      user3,
      assist3,
      user4,
      assist4,
      user5,
      assist5,
      user6,
    ],
  );

  chatStream.listen(
      (streamChatCompletion) {
        String content =
            streamChatCompletion.choices.first.delta.content.toString();

        // stdout.write(content); //  <-- Doesn't work anymore: The argument type 'String?' can't be assigned to the parameter type 
                                                    //        'String'.dartargument_type_not_assignable

        print(content); // Does this even work for streaming?

[ ... ]

Any help is appreciated. I'm sure I'm doing something wrong here.