search

weiyinwei / MMGCN

MMGCN: Multi-modal Graph Convolution Network forPersonalized Recommendation of Micro-video
283 stars 52 forks source link

Optimizer in the code is different with the paper said. #9

Closed bbjy closed 4 years ago

bbjy commented 4 years ago

Hi, @weiyinwei , your paper said 'optimizing the model with stochastic gradient descent (SGD)', but in the code, the optimizer is Adam, which is the right one you used to get the best result?

weiyinwei commented 4 years ago

Actually, for our model, SGD could achieve the best performance, but it costs too much to tune the model. In contrast, Adam gets a comparable result in a short period.