Comment by mips_avatar

The problem i've had with finetuning models is that most of the time better prompting beats finetuning