zhengbw0324
commented
1 year ago @fkuabc 您好！
对于WideDeep这类 context-aware model，要求数据存在label，MovieLens-1m原始数据集并没有直接的label数据，需要在配置文件中设置threshold参数，根据threshold生成label。例如：
https://github.com/RUCAIBox/RecBole/blob/692926f1fb4b268745402d08d11ebe59652a162f/recbole/properties/quick_start_config/context-aware_ml-100k.yaml#L1-L5
上述配置为样例数据ml-100k的配置，数据中rating大于等于4的数据label为1，否则label为0。其他数据配置信息请参考文档。
这什么玩意？ » python run_recbole.py --model=WideDeep --dataset=ml-1m
Trainable parameters: 108574 Train 0: 0%| | 0/391 [00:00<?, ?it/s] Traceback (most recent call last): File "run_recbole.py", line 25, in
run_recbole(model=args.model, dataset=args.dataset, config_file_list=config_file_list)
File "/home/RecBole/recbole/quick_start/quick_start.py", line 56, in run_recbole
best_valid_score, best_valid_result = trainer.fit(
File "/home/RecBole/recbole/trainer/trainer.py", line 335, in fit
train_loss = self._train_epoch(train_data, epoch_idx, show_progress=show_progress)
File "/home/RecBole/recbole/trainer/trainer.py", line 181, in _train_epoch
losses = loss_func(interaction)
File "/home/RecBole/recbole/model/context_aware_recommender/widedeep.py", line 65, in calculate_loss
label = interaction[self.LABEL]
File "/home/RecBole/recbole/data/interaction.py", line 131, in getitem
return self.interaction[index]
KeyError: 'label'
我还根本没用自定义配置，就用默认设置跑了一下都能出错？