Ergonomics issues with "Model/Character Settings" dialog

[brankoradovanovic-mcom]

System Info

GPT4All 2.5.2

Information

• [ ] The official example notebooks/scripts
• [ ] My own modified scripts

Reproduction

The behavior of the "Model/Character Settings" dialog can be illustrated by the following:

1. Start a chat using some model X
2. Click on the settings icon and select model X in the drop down list
3. Scroll to the bottom of the "Model/Character Settings" dialog, change some params, and close the dialog
4. Open a new chat with model Y
5. Click on the settings icon
6. The "Model/Character Settings" dialog remembers the position of the scroll bar and displays the dialog as you left it last time, which in this scenario means it's scrolled to the bottom. The name of the selected model is still X (not Y!), but this is not visible, as the name is located at the top of the dialog, out of the displayed area.

So, yesterday I spent the better part of an hour in bewilderment, as I couldn't understand why the model I used in the chat was completely resistant to changes of model params. In the end, I realized I had been inadvertently changing the parameters of one model while using another.

While this may not be a problem for some users, I'm certain many spend time experimenting and switching between different models, param settings, and chats, and in this scenario things quickly become confusing.

Expected behavior

I'd suggest the following changes:

1. When the user clicks on the Settings icon, the pre-selected model in the drop down list should be the model used in the current chat, rather than the last model selected in the Settings.
2. The "Model/Character" drop down setting should be fixed in place in the "Model Character/Settings" dialog, so that it is always visible, i.e. that it can't be scrolled away.

If I had to pick just one of the above changes, I'd pick the second one, although I believe both would be beneficial.