zlacker

[return to "Fine-tune your own Llama 2 to replace GPT-3.5/4"]
1. msp26+Tb[view] [source] 2023-09-12 17:49:20
>>kcorbi+(OP)
Do you still use few-shot prompting with a fine-tune? Or does it make little difference?
◧◩
2. selfho+jl[view] [source] 2023-09-12 18:31:34
>>msp26+Tb
In my experience, there is little need to do that. With completely unambiguous instructions that describe the exact output format, you can often get away with no examples whatsoever. Single examples might be helpful, but multi-shot prompting will be definitely unneeded (and may even harm the model's output quality).
[go to top]