What do you do if the output from Human Feedback bit is not very good? At what point it becomes futile? It's nearly impossible to align a large number of raters for tasks that aren't very black and white and there is no way to standardise the training enough to actually be repeatable across many languages and cultures. Because if you have A vs B type of decision and your group effort ends up with more or less 50:50, you don't really have anything, right?
loved it brother and keep making videos on robotics.
i am loving it and I am fully supporting you on these kinds of videos.
Awesome and really liked the way you explained.
Thanks a lot 😊
I love your videos.. please keep going and thank you for all these amazing videos
Thanks for the efforts....
What do you do if the output from Human Feedback bit is not very good? At what point it becomes futile? It's nearly impossible to align a large number of raters for tasks that aren't very black and white and there is no way to standardise the training enough to actually be repeatable across many languages and cultures. Because if you have A vs B type of decision and your group effort ends up with more or less 50:50, you don't really have anything, right?
Great and neat explanation ❤
Well presented!
Great channel, currently working on a robot arm with ai
Cool!