Fotiligner
Fotiligner
@ralgond Thanks for your attention to RecBole! In our program, when you run the run_recbole.py for the model, the process will automatically call the full_sort_predict function defined in the model,...
@njt123123 你好！ 运行run_recbole.py时，可以设定参数--model为SimpleX，然后即可运行。如果还是没法运行成功，可以关注下参数设置中的大小写。
这是多卡分布式训练时的报错，如果你使用单卡运行程序，可以将torch.distributed.barrier()注释掉；如果你使用多卡运行程序仍出现该错误，可以尝试在config文件中设置多个gpu_id
Thank you for your interest in RecBole！ We welcome you to submit PR in the github repository. You could refer to our previous model files.
@Fiveneves Hi! This is the task definition of sequential recommendation in the original paper. In RecBole, we have standardized the task data format of all models for sequential recommendation, which...
@openchao 您好！SGL文章中ssl user的loss是参考了传统Info_nce loss，在infonce中，分母部分本身也是需要包含了正样本的计算结果的。
Thanks for your attention to RecBole! We will add this function in the following update. Before that, you could manually merge the train/valid data files together as the 'part1' file.
@xxx08796 Thank you for your attention! This fix is correct, we will fix it in our following updates.
@fqsfeng 您好！我们这里训练的CPU情况都比较正常，请问可以提供详细的训练设置吗
@jinxuan0727 您好！训练结果受很多因素影响，如数据集划分方式、使用的损失函数等等，因此和原文的结果可能有一定的区别；可以提供更详细的参数设置方便我们定位问题