The structure of an article is created from key phrases. These key phrases form the basis of the future article. Based on these key phrases, Chat GPT generates H1, H2, and H3 headings, as well as the Title and Description. One line with key phrases represents one article. Key phrases are written separated by commas. You can include any number of key phrases for one article. If the number of key phrases exceeds one line, they should naturally continue onto the next line without pressing Enter. If you press Enter and start writing key phrases, it will then be considered a different article.
Additionally, you can choose images from your computer. One image will be published in one article. If you generate 10 articles, you will need to choose 10 images. However, in this case, you can also use one or several images for all the articles, and they will be duplicated in each article. To make the images unique, you can apply a watermark using the Image Watermark plugin.
2. How is the length of the article regulated?
In our plugin, the length of the article is not determined in tokens or the number of characters. Instead, it is related to the number of key phrases. The more key phrases you provide, the longer the article will be. With 3-4 key phrases, articles typically have an average length of 2000-2500 characters. Therefore, if you need a longer article, you should increase the number of key phrases for that article.
Why is it implemented this way?
We chose this approach because, with the OpenAI API, you can only specify the length of the article in tokens. However, the mechanism for controlling the article’s size in tokens is currently imperfect. Very often, Chat GPT does not finish the article if the length is specified in tokens.
The length of the article is further regulated through the selection of templates
Before generating the content, you can choose a template that will determine how the articles are generated. “Simple article” and “Quick article” are templates for standard-length articles, usually generating articles up to 5,000 characters without spaces. “Big article (Standard)” is a template for larger articles, which can be generated up to an average of 7,000-8,000 characters. “Big article (Guide)” is a template for more extensive articles with practical recommendations, with text lengths reaching up to 15,000 characters. You can also create your own templates and generate articles of various structures and lengths, but this feature is recommended for experienced users.
Why can the article length vary?
The article length depends on the chosen template, the number of key phrases, the selected GPT model, the article’s topic, and the capabilities of Chat GPT itself.
How is the article length regulated by choosing the GPT model?
In our plugin, the following models are available for content generation: GPT 3.5 with 4k tokens, GPT 3.5 with 16k tokens, GPT 4 with 8k tokens, and GPT 4 with 32k tokens. The GPT 4 model differs from the GPT 3.5 model in that it can handle more complex tasks with greater accuracy. Detailed information about their differences is provided here. In simple terms, when generating content, the GPT 4 model can produce higher-quality articles.
What does 16k or 32k mean? Which one should you choose? How is it related to the article length?
This refers to the context length. For example, if you want to use GPT 3.5, you select the “Simple article” template and make a small request with up to 4k tokens, and you receive an article with a length of up to 4k tokens. In this case, you should choose the GPT 3.5 model with 4k tokens. However, if you want to generate more extensive articles, you would select the “Big article” template, and in this case, you would send larger requests through the API. Then, you would need to choose the 16k model for GPT 3.5. The same applies to the GPT 4 models. The pricing for using these models is detailed in our FAQ.