Complete text prompt action

This action generates completions for a given prompt using OpenAI's language models. Simply provide a prompt and parameters, and the action returns one or multiple predicted completions. Use this action to autocomplete text, answer questions, and generate new content.

Complete text prompt action

Input

Field	Description
Deployment ID	Enter the deployment ID of the model you plan to use. You can find the deployment ID in Azure AI Studio > Deployment.
Prompt	The prompt to generate completions for. If a prompt is not specified, the model generates content as if from the beginning of a new document. If you plan to create responses for multiple strings (or using tokens), enter the relevant information as a datapill. For more information about the format, refer to OpenAI's documentation (opens new window).
Maximum Tokens	The maximum number of tokens to generate in the completion. The token count of your prompt plus the value here cannot exceed the model's context length. Most models have a context length of 2048 tokens (except for the newest models, such as GPT 3.5-turbo, which support 4096).
Suffix	The suffix that comes after the completion of inserted text.
Top p	Enter a value between 0 and 1 for controlling the diversity of completions. A higher value results in more varied responses. We recommend using this or temperature but not both. Learn more here (opens new window).
Temperature	Enter a value between 0 and 2 to control the randomness of completions. Higher values make the output more random, while lower values make it more focused and deterministic. We recommend using this or top p but not both. Learn more here (opens new window)
Number of completions	The number of completions to generate for each prompt.
Log probabilities	Enter a number to obtain the log probabilities on the next n (determined by this value) set of likely tokens and the chosen token. Learn more here (opens new window).
Stop phrase	A specific stop phrase that ends generation. For example, if you set the stop phrase to a period (.) the model generates text until it reaches a period, and then it stops. Use this to control the amount of text generated.
Presence penalty	A number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.
Frequency penalty	A number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood of repeating the same line verbatim.
Best of	Controls how many results are actually generated before being sent over. The number of completions cannot be less than the value input here.
Logit bias	Input JSON containing the tokens and the change in logit for each of those specific tokens. For example, you can pass {"50256": -100} to prevent the model from generating the <|endoftext|> token. Learn more here (opens new window).
User	A unique identifier representing your end-user, which can help OpenAI to monitor and detect abuse.

Output

Field Description Created The datetime stamp of when the response was generated. ID Unique identifier denoting the specific request and response that was sent over. Model The model used to generate the text completion. Choices Text The response of the model for the specified input. Finish reason The reason why the model stopped generating more text. This is often due to stop words or length. Logprobs An object containing the tokens as well as their corresponding probabilities. For example, if log probabilities was set to five, you will receive a list of the five most likely tokens. The response always contains the logprob of the sampled token, so there may be up to logprobs+1 elements in the response. Response Contains the response which OpenAI probabilistically considers to be the ideal selection. Usage Prompt tokens The number of tokens used by the prompt. Completions tokens The number of tokens used for the completions of text. Total tokens The total number of tokens used by the prompt and response.