This is the first time I've used an LLM as small as this... what is it for?
It is bad, hallucinate a lot... I'm just curious.. I can't see how small models like this can be useful, it struggle with a simple conversation.
Its funny 😂
can't speak for or against this specific model - small models tend to be very sensitive to configuration, input, template, use... everything. most very small models are somehow specialised, i.e. need a special prompt, are only meant to answer specific questions/topics, or do specific tasks such as translation. and a lot are experiemnts, to see how effective certain methods are before wasting more money on bigger models.
also, this model is comparatively old, things have gone a long way. for example, lfm2 models already perform quite well at general conversation.
the smallest reasonably usable model ive tested is qwen3 1.7b and that one i like to set up as an agent, making it analyze something and output a very specific yes/no format. it can do this very very quickly and i use it to aid larger model