ChatGPT Assistant is a Joomla module that includes a chat button in your page, introducing the ability for your users to discuss with ChatGPT directly from your website. This DOES NOT mean that ChatGPT Assistant will automatically adjust ChatGPT to answer questions only related to your website. You are responsible for that, by leveraging the initial model instruction option. For more information about ChatGPT Assistant features, please visit the download page.

No. ChatGPT Assistant is a Joomla! extension and can only be used within the Joomla! CMS.

Please refer to the ChatGPT Assistant download page, on the right.

Sure. The module comes with a language file that allows translation. The ChatGPT responses however, depend on the user input.

No. Without any context, ChatGPT Assistant will behave more or less like the original ChatGPT. In order to customize the assistant responses, you will need to provide relevant context, using the initial model instruction. Currently, only the Chat Completions API is implemented, which does not allow model fine-tuning. If in the future, model fine-tuning is implemented for the Chat Completions API, ChatGPT Assistant will be compatible with it, and allow for further customization of the assistant's responses.

No. At the moment, OpenAPI has not allowed fine-tuning for the Chat Completions API. If in the future, model fine-tuning is implemented for the Chat Completions API, ChatGPT Assistant will be compatible with it, and allow for further customization of the assistant's responses.

To install ChatGPT Assistant, you follow the usual steps, as with installing any other Joomla! extension.

Navigate to Extensions > Extension Manager, and under the tab "Upload Package File", you select the zip archive you downloaded from Mavrosxristoforos.com, and click "Upload & Install".

Wait until the package file gets installed and you should see a completion message. You are ready to go.

After installation, the first thing you should do, is enable the chat button, by publishing the module in any desired module position. You also need to write your OpenAI API key, in the module options. You can get an API key here.

In order to provide relevant context and allow for certain customization of the model's responses, you can use the Initial Model Instruction field. There is virtually no limit to what context you can provide there, however ChatGPT may sometimes produce inaccurate responses, and there is also a price per token of input, so you may want to perform a lot of tests before choosing what is best for your use case.

This option can be used to give the model some initial instructions about its scope. For example, you can write something like "You are a helpful Joomla expert with a limit of 5 messages. You must only provide answers related to our Joomla extensions and clients' Joomla websites. You politely decline questions irrelevant to Joomla or our Joomla extensions. If someone needs further assistance, kindly ask them to use the website's contact form." Having this message will help the model mostly reply about Joomla, but this is still not guaranteed.

You can view our Youtube video, or read about it here.

Please configure your OpenAI API key in the module options. You can get an API key here.

This means that ChatGPT took longer than expected to respond. If this is common, you may increase the time limit in the Advanced tab of the module, using the Curl Opt Timeout field.