Create Chat Completion API and Create Image API Configuration Settings
The configuration setting for the OpenAI integration features included in the eLearning Magic Toolkit plugin can be found in your WP-Admin Settings>eLM Toolkit screen, as soon as the plugin has been activated and licence key registered. This page will describe the impact that each setting will have on the response results you can expect back when using either API:
Create Chat Completion API
Model – You can select from a number of different models which are made available by OpenAI to use as an API solution. Each model has its own limitation and costs associated with their use, so make sure that you make your selection wisely after reviewing the current pricing table.
Temperature – is a number between 0 and 2, with a default value of 1 or 0.7 depending on the model you choose. The temperature value is typically used to control the randomness of the output. If you need for responses to be more direct and certain, try dropping the number below 0.7.
Max Tokens – Cost for using the OpenAI platform is calculated based on token spend. 1 token is typically classed as 4 characters, and so equates to roughly 3/4 of a standard word used per prompt. A token cost is generated for both the request and response text that is generated by the AI model. You can use the Max Tokens value setting in order to control the top limit that you wish for any AI request to spend from your OpenAI platform account. This can be useful to ensure that you are able to maintain a monthly budget spending/usage limit for example.
Presence Penalty – Typically managed alongside the Frequency Penalty, the Presence Penalty value is a number between -2.0 and +2.0. A higher Presence Penalty value will result in the model being more likely to generate tokens that have not yet been included in the generated text, thus promoting the generation of new ideas and diversity within responses.
Frequency Penalty – The Frequency Penalty value is also a number set between -2.0 and +2.0 A higher Frequency Penalty value will result in the model being more conservative in its use of repeated tokens, thus reducing the amount of repeated phrases.
Create Image API
Image Size – The available image sized which the Create Image (Dall-E 2) model can produce are either 256×256, 512×512, or 1024×1024. As the image size request goes up, so too does the cost of generation per image. You can make your choice based on your budget and what works best for the resolution of image needed to appear in your learning experience.
Read the rest of our Knowledge Base pages to make the most of eLearning Magic Toolkit.