[QUESTION] Q about sequential models about recbole

[ShanleiMu]

Please refer to #614.

[chenyushuo]

We always do data augmentation in the sequential model. You can see the following code for details:

RecBole/recbole/data/dataset/sequential_dataset.py

Lines 40 to 94 in 90d5c3f

	def prepare_data_augmentation(self):
	"""Augmentation processing for sequential dataset.
	E.g., ``u1`` has purchase sequence ``<i1, i2, i3, i4>``,
	then after augmentation, we will generate three cases.
	``u1, <i1> | i2``
	(Which means given user_id ``u1`` and item_seq ``<i1>``,
	we need to predict the next item ``i2``.)
	The other cases are below:
	``u1, <i1, i2> | i3``
	``u1, <i1, i2, i3> | i4``
	Returns:
	Tuple of ``self.uid_list``, ``self.item_list_index``,
	``self.target_index``, ``self.item_list_length``.
	See :class:`SequentialDataset`'s attributes for details.
	Note:
	Actually, we do not realy generate these new item sequences.
	One user's item sequence is stored only once in memory.
	We store the index (slice) of each item sequence after augmentation,
	which saves memory and accelerates a lot.
	"""
	self.logger.debug('prepare_data_augmentation')
	if hasattr(self, 'uid_list'):
	return self.uid_list, self.item_list_index, self.target_index, self.item_list_length
	self._check_field('uid_field', 'time_field')
	max_item_list_len = self.config['MAX_ITEM_LIST_LENGTH']
	self.sort(by=[self.uid_field, self.time_field], ascending=True)
	last_uid = None
	uid_list, item_list_index, target_index, item_list_length = [], [], [], []
	seq_start = 0
	for i, uid in enumerate(self.inter_feat[self.uid_field].values):
	if last_uid != uid:
	last_uid = uid
	seq_start = i
	else:
	if i - seq_start > max_item_list_len:
	seq_start += 1
	uid_list.append(uid)
	item_list_index.append(slice(seq_start, i))
	target_index.append(i)
	item_list_length.append(i - seq_start)
	self.uid_list = np.array(uid_list)
	self.item_list_index = np.array(item_list_index)
	self.target_index = np.array(target_index)
	self.item_list_length = np.array(item_list_length)
	return self.uid_list, self.item_list_index, self.target_index, self.item_list_length

[mayaKaplansky]

Thats perfect thank you so much!!

[cramraj8]

@chenyushuo

When I run the run_recbole for a GRU4Rec in sequential setting, the data augmentation were never run through. I tried to put print statements in the scripts to see if the data_augmentation function is called (

RecBole/recbole/data/dataset/sequential_dataset.py

Line 49 in 0c7197b

self.data_augmentation()

), but it never did.

And I am getting the below error because of that

"/home/xxx/xxx/RecBole/RecBole/recbole/model/sequential_recommender/gru4rec.py", line 84, in forward
    seq_output = self.gather_indexes(gru_output, item_seq_len - 1)
  File "/home/xxx/xxx/RecBole/RecBole/recbole/model/abstract_recommender.py", line 174, in gather_indexes
    output_tensor = output.gather(dim=1, index=gather_index)
RuntimeError: index 122111 is out of bounds for dimension 1 with size 53

Any ideas how I can solve it or am I missing any config statements ?