Can I have a example of your LlamaFactory tuning yaml template?
#4
by
Neroism8422
- opened
I just had some trouble in reproducing the result, seems like the model didn't get the format of Reactants.Reactants or Reactants.Reactants>Product, lead to reaction based tasks has low performance, only forward reaction has the performance in paper.
I was in a really really bad situation in my own research, may I have some kind of tips when you train your model? I really need it, pls... like how you handle this kind of format issue, just feed enough data and tune the parameter in a right position? And are there any thing that I should pay attention in tuning these kind of Chemistry task? The prompt, the data structure something like that?