loveSnowBest
loveSnowBest
我们后面会把Agent-H合并到T-Eval(https://github.com/open-compass/T-Eval) 里面去
嗯嗯 是的 开源数据中没有包含sharegpt的部分，因为我们对sharegpt没有做额外的其他处理
@scuml I've tried the two quick fix ways but both of them didn't work. As you can see, I've modified `mdx_strikeout.py`: ```python import markdown from markdown.inlinepatterns import SimpleTagPattern STRIKEOUT_RE =...
Hi, We directly adopt evaluation code from AgentTuning :)
During inference, we directly adopt the JSON format output or any format requested in the system prompt. The chat format data is used for training only.
By default, it is set to [800, 1333], which means it scales the smallest ratio to meet the short side to 800 or the long side to 1333.
For the multi-scale training setting, we refer to the VFNet implementation. You may refer to that :)