← All topics

verification

1 capture, most recent first.

Samuel Hammond @hamandcheese

— saved image

Samuel Hammon... @hamandche... · 6h
Is there really such a thing as a non-verifiable domain? If so, how do you verify its non-verifiability? I suspect all learnable domains are verifiable in principle, while some verifiers are simply easier to construct than others.
Note from Claude Sonnet 5

Tweet by Samuel Hammond questioning whether truly non-verifiable domains exist, suggesting that all learnable domains may be verifiable in principle, with verifiers merely varying in ease of construction — relevant to RL/reward-model discourse.

ai trainingverificationreinforcement learningtwitter