IronBeliever / CaR

Clustering and Ranking: Diversity-preserved Instruction Selection through Expert-aligned Quality Estimation
49 stars 1 forks source link

请问Table 2中是使用什么基准模型进行比较的 #3

Open 4real3000 opened 2 months ago

4real3000 commented 2 months ago

WX20240408-145451@2x 请问在表2中是使用什么模型和Alpaca-PandaLM、Alpaca-cleaned、Vicuna、Alpaca、Alpagasus、AlpaCaR这几个模型进行对比产生WS、WR、QS这三个结果的。 您的文章对我很有帮助,期待您的回复,谢谢!

IronBeliever commented 2 months ago

感谢您对这篇工作的关注!

主实验部分的baseline使用的是数据集的reference response。\ 我们也注意到论文确实没有说明这一点,即将在新的版本中修正😀