I've been doing meta prompting, result isn't so good for small models, it seems (maybe I'm wrong) these small models especially Quantized models they tend like short conscience prompts so short prompt with LoRa in my opinion is the best way rather than complicated or too sophisticated prompts which are favorable by frontier models, most big models love that kind complete long prompt.
![BLACK BAG - Official Trailer [HD] - Only in Theaters March 14](http://i.ytimg.com/vi/Du0Xp8WX_7I/mqdefault.jpg)
I've been doing meta prompting, result isn't so good for small models, it seems (maybe I'm wrong) these small models especially Quantized models they tend like short conscience prompts so short prompt with LoRa in my opinion is the best way rather than complicated or too sophisticated prompts which are favorable by frontier models, most big models love that kind complete long prompt.