Hi, thanks for your insightful work!
I find the results in the paper a bit hard to follow. Could you share your training script and augmented dataset used if possible, so that I can confirm if I missed something.
Also, does the evaluation taken on the origin IFEval repo? Could you share the prompt for evaluation response generation?
Hi, thanks for your insightful work!
I find the results in the paper a bit hard to follow. Could you share your training script and augmented dataset used if possible, so that I can confirm if I missed something.
Also, does the evaluation taken on the origin IFEval repo? Could you share the prompt for evaluation response generation?