Description
The OpenAICompletion component is a component that integrates with the AI features provided by OpenAI. OpenAICompletion returns completions for given prompts. For chat-based functions, please use the OpenAI Chat.
※ To use this OpenAICompletion component, you need to obtain an API Key in advance to call OpenAI's AI function. For how to get an API Key on OpenAI's site, please refer to OpenAI API Key Setup.
Component Properties
| Property Name | Description | Example |
|---|---|---|
| Basic Properties | Specify the basic properties of the 'OpenAI' category | |
| API Key | API Key ※1 | |
| Organization Id | Organization ID ※1 | |
| User | Username | |
| Model | Choose or manually input the model name to use | Text Davinci 003 Text Davinci 002 Text Curie 001 Text Babbage 001 Text Ada 001 |
| Temperature | Specify the sampling temperature between 0 and 1 | |
| Prompt | Prompt | |
| Suffix | Suffix | |
| Max Tokens | The maximum number of tokens to be generated | |
| Top-P | Specify nucleus sampling between 0 and 1. Default is 1 | |
| N | The number of answers generated per prompt | |
| Log Probabilities | Include the log probabilities of the most likely token and the selected token | |
| Echo | Add the prompt to the answer | |
| Stop | String to stop token generation | |
| Presence Penalty | Specify whether to use a word that has already appeared again, from -2.0 to 2.0. Default is 0 | |
| Frequency Penalty | Specify whether to give a penalty the more often it appears, from -2.0 to 2.0. Default is 0 | |
| Best Of | Control the number of candidate answers. You specify the number to return, but best_of must be greater than N | |
| Logit Bias | Parameter to change the likelihood of a specified token appearing |
※ Please refer to the OpenAI API Reference for details on each OpenAI-related property.
※1 To use the OpenAICompletion component, you need to obtain an API key. For how to get an API key, please refer to OpenAI API Key Setup.
Output Payload
| Output Payload | Description |
|---|---|
cv.Payload | It stores the response from the OpenAI Completion API as is |