8/6/2023 0 Comments Java chatbot apiNow that you know what ChatGPT is, it’s time to start building your own chat solutions. When I ask the assistant to format the list as a table, it leverages the previous messages in the chat history to source the information it needs to populate the table. The chat history contains the list of initial recommendations. It uses the previous messages its history to ask for clarification on the Hudson Valley area and intensity. ![]() When I ask for Hudson Valley hike recommendations, as instructed by the system message, the assistant follows up with questions to provide more personalized recommendations. Using the same hiking assistant example, here are a few examples where chat history is being used as context for additional follow-ups: With ChatGPT you can leverage the chat history as additional context. Therefore, if you want to ask follow-up or additional questions you have to find a way to embed it into the context of a prompt. Generally, when working with GPT-3 models the prompts and responses are one-off. One of the things you’ll notice is that the responses and interactions with the model are guided not only by the user messages, but also by the initial system message. The responses generated by the model are on the left and these have the assistant role. The messages on the right are the messages I compose. The following screenshot is what a sample chat with a ChatGPT powered digital hiking assistant might look like. You will also share an interesting fact about the local nature on the hikes when making a recommendation. You will then provide three suggestions for nearby hikes that vary in length after you get that information.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |