| Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
While it shouldn't take very long, this should nonetheless
avoid race conditions on the task assignment logic. Following
the discussion here: https://stackoverflow.com/q/18213619
|
|
This commit changes the task assignment to randomly
assign tasks that need more annotations with a bias
towards those that are close to being annotated with
our desired number of annotations. Users are further
allowed to keep annotating datasets if they want to.
|
|
|
|
|
|
|
|
|
|
With the demo in place, we're rewriting the
task assignment flow such that users only
get a task assigned when:
1. They finish the demo
2. They finish a task
3. They login again.
This way we can better balance the datasets
and we won't have datasets that don't get
enough annotations because some users didn't
finish tasks they were assigned.
|
|
Also, add automatic assign on user signup
|