I have a use case where I need to set the order of the data rows for labelers to go through specific groups of text data sequentially.
I understand that I could create a batch per group (or set data row priority for an existing batch) and assign a different
priority to each data group. However, I have scenarios with > 20 groups but can only increment the
priority from 1 to 5.
Are there any other ways to set the order of data rows in a project?
Thank you for posting!
Only the priorities per data rows (or Labeling Parameter Override) can support more than 5 levels of priorities.
In your example, what do you mean by “group”? Do you need to set priorities between 1 and 5 for all the data rows of a set of batches?
Thanks for your reply @PaulN,
The documentation on set data row priority says that the
priority argument of the LPO must be between
Essentially I need to set the priorities for all the data rows in a project between say
30 regardless of whether I queue the data in one big batch (and use LPOs) or in 30 batches with different priorities
Setting the data row priorities using LPOs seems to work, thanks for the suggestion.
I think this part of the documentation is misleading:
Each override has three values: the index value of the target data row, the new priority, and the number of rows affected. (All values must be positive integers that match the range of the list. The priority must be between
Thanks @tw974 , there seems to be a typo here. Although Labelbox recommends using priorities between 1 and 5 (like for batches), LPOs offer more flexibiliy. You should be able to use an integer for the priority value.
@PaulN should I be worried about the warning below?
LabelingParameterOverrides are deprecated for new projects, and will eventually be removed completely. Prefer to use batch based queuing with priority & consensus number of labels instead.
Does that mean that batch based priority values won’t be constrained to 1-5 anymore?
This is just a warning. Please disregard it.