In the processes.py file, line 572, when train_log_interval is not provided it is set to len(training_dataset). Hence, the first part of the if-test is only True if iteration is a multiple of len(training_dataset).
However, pbar.total is also set to len(training_dataset) and thus the second test can never pass when iteration is equal to len(training_dataset), which is when the if-test should actually be succeeding. Should this not be:
if iteration % train_log_interval == 0 and pbar.total >= iteration:
Instead of:
if iteration % train_log_interval == 0 and pbar.total != iteration:
2
In the utils.py file, line 106, when the dataset does not explicitly provide event ids, all event ids should be selected.
However, event_ids is set to None and therefore line 106 results in an AttributeError:
events = events[[i for i in range(len(events)) if events[i, -1] in event_ids.keys()], :]
1
In the processes.py file, line 572, when train_log_interval is not provided it is set to len(training_dataset). Hence, the first part of the if-test is only True if iteration is a multiple of len(training_dataset). However, pbar.total is also set to len(training_dataset) and thus the second test can never pass when iteration is equal to len(training_dataset), which is when the if-test should actually be succeeding. Should this not be:
if iteration % train_log_interval == 0 and pbar.total >= iteration:Instead of:
if iteration % train_log_interval == 0 and pbar.total != iteration:2
In the utils.py file, line 106, when the dataset does not explicitly provide event ids, all event ids should be selected. However, event_ids is set to None and therefore line 106 results in an AttributeError:
events = events[[i for i in range(len(events)) if events[i, -1] in event_ids.keys()], :]