Neural / neural sparse - any way for it to understand negatives?

Plugins Machine Learning
1

Versions Latest

Is there any way to make the neural or neural_sparse searches understand negative prompts? I spun up a test server and just loaded their recommended model for both and they work really well when I test normal postive questions: “Show me all the british cheeses” but if i try it with a negative it ignores it (well it brings blue cheese to the top): “Show me all the british cheeses apart from blue”

I understand why its doing it but is there away for the model to pull it apart and take the negatives and positives and perform different types of searches?

Thanks,

3

I am guessing something like this:

Going to see if I can get it running with OLLAMA