Re:Gopher – A 280B parameter language model
by singularity2001 on Wednesday December 22, 2021 @07:17AM
(#29648255)
How would one train an additional output neuron for 'confidence'?
Re:Gopher – A 280B parameter language model
by paunchy on Wednesday December 08, 2021 @12:58PM
(#29488071)
I think you're missing the point. That section was to show that the model is sometimes wrong and lacks the self-awareness to be uncertain about that wrong answer.
They're transparently providing an example where their product doesn't work well. Find me another product, even an OSS project that does the same on their landing page.
Re:Gopher – A 280B parameter language model
by canjobear on Wednesday December 08, 2021 @03:42PM
(#29490216)
Ha, you're right. Whoops!
Re:Gopher – A 280B parameter language model
by BillinghamJ on Wednesday December 08, 2021 @12:31PM
(#29487700)
Is it not in that section specifically because they're showing failure situations?
Not a good look to have an error on the landing page for your product.