Open ralgond opened 1 year ago
@ralgond in short, you could sample some candidate items, scoring by the model, select topk items with higher scores, similar details shown in https://github.com/pmixer/SASRec.pytorch/blob/master/utils.py#L140 , still, the repo is for recsys research use, you may need to do some engineering work if hope to train a model and deploy it for online inference for production.
thank you!