CHAML 除 0 问题

[Wonderdch]

CHAML 严格按照 readme 操作后，在最后一步执行 run.sh 命令后报错“除 0”。

详细错误代码是：

(CHAML) dch@gpuadmin-SYS-7048GR-TR:~/source/CHAML$ sh run.sh
06/10 11:21:02 - INFO -   method is meta. HARD_TASK: True, HARD_USER: True, CURRICULUM: True, PACING_FUNCTION: ssp, PER_TEST_LOG: 2500, PATIENCE: 2
06/10 11:21:02 - INFO -   curriculum is: [0, 1, 2, 3, 4, 5, 6, 7]
06/10 11:21:02 - INFO -   Got config from config/config-chaml.json
{'update_lr': 0.001, 'meta_lr': 0.001, 'update_step': 1, 'update_step_test': 1, 'task_batch_size': 4, 'train_qry_batch_size': 512, 'max_train_steps': 100000, 'few_num': 512, 'num_poi_types': 230, 'num_time': 25, 'embed_dim': 50, 'poiid_dim': 50, 'mlp_hidden': 300, 'local_fix_var': 1, 'global_fix_var': 1, 'sample_batch_size': 1024, 'test_task_batch_size': 1, 'num_epoch': 10, 'with_cont_feat': True}
<model.meta.Meta object at 0x7fbc1dded7d8>
06/10 11:21:02 - INFO -   Total trainable tensors: 225653
06/10 11:21:06 - INFO -   Loaded all the data pickles!
/home/dch/source/CHAML/utils/metadataset.py:57: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  final_pos_samples.append(np.array([user_id, hist, pos_candi, label]))
/home/dch/source/CHAML/utils/metadataset.py:58: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  final_neg_samples.append(np.array([user_id, hist, neg_candi, 0]))
/home/common/anaconda3/envs/dch-CHAML/lib/python3.6/site-packages/paddle/fluid/dygraph/math_op_patch.py:165: RuntimeWarning: divide by zero encountered in double_scalars
  return _scalar_elementwise_op_(var, 1.0 / value, 0.0)
Traceback (most recent call last):
  File "main.py", line 374, in <module>
    main_meta(meta_path, root_path, id_emb_path)
  File "main.py", line 283, in main_meta
    'stage1', CURRICULUM, HARD_TASK, batch_id=batch_id)
  File "main.py", line 222, in one_meta_training_step
    poiid_embs=poiid_embs, cont_feat_scalers=cont_feat_scalers)
  File "/home/common/anaconda3/envs/dch-CHAML/lib/python3.6/site-packages/paddle/fluid/dygraph/layers.py", line 891, in __call__
    outputs = self.forward(*inputs, **kwargs)
  File "/home/dch/source/CHAML/model/meta.py", line 76, in forward
    vars=None, scaler=scaler)
  File "/home/common/anaconda3/envs/dch-CHAML/lib/python3.6/site-packages/paddle/fluid/dygraph/layers.py", line 891, in __call__
    outputs = self.forward(*inputs, **kwargs)
  File "/home/dch/source/CHAML/model/learner.py", line 133, in forward
    hist_embed, mask)
  File "/home/dch/source/CHAML/model/learner.py", line 73, in attention
    score = paddle.where(mask==1, wall, score)
  File "/home/common/anaconda3/envs/dch-CHAML/lib/python3.6/site-packages/paddle/fluid/dygraph/math_op_patch.py", line 238, in __impl__
    return math_op(self, other_var, 'axis', axis)
RuntimeError: (NotFound) Operator equal does not have kernel for data_type[bool]:data_layout[ANY_LAYOUT]:place[CPUPlace]:library_type[PLAIN].
  [Hint: Expected kernel_iter != kernels.end(), but received kernel_iter == kernels.end().] (at /paddle/paddle/fluid/imperative/prepared_operator.cc:127)
  [operator < equal > error]