I'm trying to compute movielens, after downloading movielens 20M, computing the data transformation, when I run the recommender:
python run.py --gpu_ids 0 --path_to_train_data movielens/train --path_to_eval_data movielens/val --hidden_layers 512,512,1024 --non_linearity_type selu --batch_size 128 --logdir model_movielens --drop_prob 0.8 --optimizer momentum --lr 0.005 --weight_decay 0 --aug_step 1 --noise_prob 0 --num_epochs 1 --summary_frequency 1000
I get an error:

Loading training data
Data loaded
Total items found: 138493
Vector dim: 20668
Loading eval data
Traceback (most recent call last):
  File "run.py", line 236, in <module>
    main()
  File "run.py", line 116, in main
    item_id_map=data_layer.itemIdMap)
  File "/home/hoaphumanoid/notebooks/repos/DeepRecommender/reco_encoder/data/input_layer.py", line 48, in __init__
    value = minor_map[int(parts[self._minor_ind])]
KeyError: 18223

It looks that it is trying to find a key in the evaluation dataset that is not in the training dataset?