crashes at launch

[patriot0pirate]

Bug Report

When I double-click the desktop icon, the program flashes on the screen then vanishes.

Steps to Reproduce

1. I downloaded the gpt4all-installer-linux.run file
2. Using the wizard, I installed gpt4all
3. The program ran fine
4. I chose and installed a model compatible with local docs
5. loaded the model and chose the folder containing the docs
6. The process completed
7. gpt4all crashed
8. I attempted to load the program again by clicking the desktop icon but the program flashes on the screen and immediately crashes
9. I uninstalled and reinstalled the program
10. I rebooted the computer. Nothing changed
11. When I double click the desktop icon, the program flashes on the screen and crashes

Expected Behavior

I expected the program to run smoothly

Your Environment

• GPT4All version: 2.7.3
• Operating System: ubunti 22.04
• Chat model used (if applicable): all-MiniLM-L6-v2-f16.gguf

[cosmic-snow]

A few things for your information:

• all-MiniLM-L6-v2-f16.gguf is not a chat model, but a model that enables the LocalDocs feature. These are distinct categories, you need one of both for LocalDocs to work. (E.g. for a chat model, try a Mistral one.)

• There must be some settings that weren't cleaned up when uninstalling. That's probably something that should be addressed in one form or another. But I guess unconditionally deleting the settings wouldn't be great, either.

• For the chat GUI, you can find the .ini configuration file in ~/.config/nomic.ai/. If you ever want to purge everything, there are also some files being put into ~/.local/share/nomic.ai/GPT4All/ and possibly ~/.cache/ (not sure, and not sure if there are other locations).

• For your case, I think it should be enough to delete the .ini file in the config folder mentioned above.

What happens here is probably that it loads or tries to load a model that is not meant to be loaded as a chat model, then crashes.

Thanks for the details you provided, but I'm not entirely sure about step 4. Can you also say from what screen you downloaded all-MiniLM-L6-v2-f16.gguf, or what steps you took in more detail? That is, did you just download it from the initial download list popup?

[cebtenzzre]

What CPU do you have? And in order to accurately diagnose the crash, we're going to need a backtrace.

1. Download gpt4all-installer-linux-v2.7.3-debug.run from here
2. Uninstall your existing GPT4All and install the debug version
3. Install gdb if you don't already have it
4. Run gdb ~/gpt4all/bin/chat (assuming you installed to the default location)
5. run to start it
6. If it crashes:
   1. set logging on
   2. set logging file backtrace.log
   3. thread apply all bt
   4. Upload backtrace.log here.

[cebtenzzre]

Closing due to lack of response.