The model Deepseek-Math-7B-base cannot reproduce the ratings in the paper #12
Description
First of all, I would like to express my sincere gratitude to the authors for their wonderful work.
However，I couldn't reproduce the evaluation results related to deepseek using the script you provided.When I was testing, the base model of deepseek basically had no ability to answer mathematical questions.Could the author provide more details about the evaluation related to deepseek-math-base？
I'm really looking forward to the author's reply. Thank you very much!
@Wyyyb