Selected model is not preserved when clicking on "New Chat"

[brankoradovanovic-mcom]

System Info

GPT4All 2.4.4 AVX Windows 10 x64

Information

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

Related Components

• [ ] backend
• [ ] bindings
• [ ] python-bindings
• [X] chat-ui
• [ ] models
• [ ] circleci
• [ ] docker
• [ ] api

Reproduction

I have two installed models, mpt-7b-chat and gpt4all-l13b-snoozy.

1. Do some chatting with gpt4all-l13b-snoozy (it is the model selected in the drop down list)
2. Click on "New Chat"
3. mpt-7b-chat is loaded and displayed in the drop down list
4. That's not the model you wanted, so you select gpt4all-l13b-snoozy again

This is the process that takes some time, as mpt-7b-chat is loaded, and then gpt4all-l13b-snoozy is loaded again

Expected behavior

1. Do some chatting with gpt4all-l13b-snoozy (it is the model selected in the drop down list)
2. Click on "New Chat"
3. gpt4all-l13b-snoozy is still displayed in the drop down list. mpt-7b-chat is not loaded because there's no need for it

Generally speaking, models should not be loaded unless it's necessary. I'm currently stuck with a laptop with a 200 Mb/s SSD, so for some users this is a 10 sec+ operation.