ValueError: `training_neg_sample_num` should be 0 if inter_feat have label_field [label].

I am getting the following error for LightGCN

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-14-0b2a29229228> in <module>
     47 
     48 # dataset splitting
---> 49 train_data, valid_data, test_data = data_preparation(config, dataset)
     50 
     51 # model loading and initialization

c:\users\laxmi\downloads\recbole\recbole\data\utils.py in data_preparation(config, dataset, save)
    101     if config['training_neg_sample_num']:
    102         if dataset.label_field in dataset.inter_feat:
--> 103             raise ValueError(f'`training_neg_sample_num` should be 0 '
    104                              f'if inter_feat have label_field [{dataset.label_field}].')
    105         train_distribution = config['training_neg_sample_distribution'] or 'uniform'

ValueError: `training_neg_sample_num` should be 0 if inter_feat have label_field [label].

If I set training_neg_sample_num=0 then I get another error saying

c:\users\laxmi\downloads\recbole\recbole\model\general_recommender\lightgcn.py in calculate_loss(self, interaction)
    136         user = interaction[self.USER_ID]
    137         pos_item = interaction[self.ITEM_ID]
--> 138         neg_item = interaction[self.NEG_ITEM_ID]
    139 
    140         user_all_embeddings, item_all_embeddings = self.forward()

c:\users\laxmi\downloads\recbole\recbole\data\interaction.py in __getitem__(self, index)
    103     def __getitem__(self, index):
    104         if isinstance(index, str):
--> 105             return self.interaction[index]
    106         else:
    107             ret = {}

KeyError: 'neg_pid'

Any idea, how to remove this error?

[2017pxy]

@laxmimerit Yes! You are right, the process of converting token to id is called remap, which is done in dataset.py, and I must admit that it may bring some confusion to new comers.

[chenyushuo]

Can you offer the full output of error? I’m confused, too. And you said that it will raise error when you pass some big number of uid, can you offer train_data.dataset.user_num and train_data.dataset.item_num?