One of the most important parts of the document editor is the AI provider configuration. Here you will control all aspects of the AI and set the parameters with which you will start creating content.
You will find it on the left side of the screen as shown in the following image:
If you are creating content with your phone, you can access the settings from the following buttons:
Default settings #
At the top of the configuration options, you will find two buttons that will allow you to set a default configuration:
Once you have set your preferred settings, you can save them with the Save button. If at some point you modify them to experiment, you can reload them with the Load button.
Provider selection #
In the next field, you will find the different providers available, which you can choose according to your needs, as well as quickly swap them during the content creation process.
If you select a provider for which you did not enter an API key, the generate button will not be available, and instead you will have a link to the configuration page.
AI Parameters #
Below you will find the different models and parameters available for the AI according to the selected provider. You can change them using the slider or by entering the value directly.
These parameters are specific to each provider. In that sense, there is no right or wrong configuration; you must define them according to your needs. You can see their descriptions by hovering over the parameter name.
Generally speaking, we can give you some recommendations for your optimal settings.
- For creative texts (marketing) it is recommended to use high temperature values. For factual texts (essays, biographies, etc.), low temperature values are recommended.
- The frequency penalty is useful when the technology starts to become repetitive.
- The presence penalty increases the likelihood that the AI will write about a new topic.
- For experimentation, you can try increasing the number of results to be obtained, this way you will have a varied view of the results and save on the use of tokens, since in a single call you will get different results. (This does not apply to Muse).
To know specifically the parameters of each provider, we leave you a link to their official documentation.
- Models: https://beta.openai.com/docs/models/gpt-3
- Parameters: https://beta.openai.com/docs/api-reference/completions/create
- Models: https://studio.ai21.com/docs/jurassic1-language-models/
- Parameters: https://studio.ai21.com/docs/api/#payload
- Models: https://muse-docs.lighton.ai/api/models
- Parameters: https://muse-docs.lighton.ai/api/endpoints/create/
With the generate button, the content will start to be created according to the set parameters and will be included in the text editor. You can also find this generation in the AI History section. You can use the keyboard shortcut Control + J (Command + J) as an alternative.