search

RUCAIBox / NCL

[WWW'22] Official PyTorch implementation for "Improving Graph Collaborative Filtering with Neighborhood-enriched Contrastive Learning".
119 stars 20 forks source link

parameter setting #27

Closed Iwillcome closed 2 years ago

Iwillcome commented 2 years ago

There are some problem about parameter setting. In the paper,the range of regularization λ3 did't introducted. Meantime, i find it is different for different dataset in yaml file, for example, the λ3 is 1e-6 and 1e-4 in amazon-books and yelp datasets respectively. Other paramter setting also different the description in paper. So, I want know the range of λ3 and the current λ3 whether or notthe best value in experiment.

linzihan-backforward commented 2 years ago

Hi! λ3 is the coefficient of L2 regularization which is not the main contribution of our model, so we omit the range in our paper. The current λ3 in our yaml file is the best value and the range of λ3 is [1e-3, 1e-4, 1e-5, 1e-6].

Iwillcome commented 2 years ago

Thanks for you response. now i understand the λ3 , lr and other parameters are the best value for NCL. Thanks you again.