Crash on loading embedded model.

[77ZaRR77]

At current time, the download list of AI models shows aswell embedded ai models which are seems not supported.

1. Download from gpt4all an ai model named bge-small-en-v1.5-gguf
2. Restart programm since it won't appear on list first.
3. Choose the ai model - crash

Expected Behavior

An epic AI talks to you back telepatically and give everything what you wish (virtually only)

Environment

• GPT4ALL 2.7.2
• Win 10 64bit.
• EMBEDDED CompendiumLabs/bge-small-en-v1.5-gguf

[manyoso]

Here is the stack...

1. We should find some way to discern that it is an embedding model and list it as such clearly in the GUI download dialog
2. We should not allow the selection of this model in the GUI after it has been downloaded ... rather, this model can only be used to power localdocs
3. Regardless, we should not crash even if the model is attempted to load in the main GUI

There are tags in huggingface for 'text-generation' and we can do a better job of listing these tags in the download dialog or filtering on them. other than that we could also potentially find out by the 'model_type' tag that lists the general family and in this case it would be a bert based model which is synonymous with embedding models.