You need to train this model more or its basically useless
#2
by
rombodawg
- opened
You do realize training a Lora on 1 epoch basically does nothing right?
3 epochs is MINIMUM for a good Lora, and even then I recommend at least 4-5 to get the most effective finetune when using Lora.
Essentially all you did with this model is teach it a style of inference, like a character card would accomplish about as much work as this finetune.
I'm not trying to dog on you too hard, or be mean, I'm just trying to pass on some constructive criticism.