Activities

The following table lists the activities that are included in this package.

Activity	Description
Generate Text Completion	Given a prompt and other inputs such as the Model ID, etc., this activity generates a text completion response. This functionality makes use of legacy GPT-3 text generation models such as Curie, Ada and Babbage. See more about the differences between models on the Chat Completions vs. Completions page. Also, you can read more on text completion from the Learn how to generate or manipulate text page.
Generate Chat Completion	Given a user prompt and additional options like instruction (that may include conversation history), this activity generates a text completion response. This activity defaults to the GPT-3.5-Turbo model, but also allows access to GPT-4. Both of these models are optimized for chat, hence the name of the activity. You can read more about the chat functionality from the Learn how to work with the GPT-35-Turbo and GPT-4 models page. While giving users the ability to add conversation history, this activity does not actually keep previous requests and responses from the model in memory. It is state-less. Each Chat activity would require the user to provide the full dialogue history as context in the prompt.
Create Embeddings	Get a vector representation of a given input that can be easily consumed by machine learning models and other algorithms. This activity is intended to be used in conjunction with a vector database and/or the embedded model for thext and chat completion.