← Back to context Comment by danielmarkbruce 2 days ago It's not an estimation of something. It's a policy. 2 comments danielmarkbruce Reply gwerbin 1 day ago Sure, you're right.But it's a policy learned from a next-token prediction task. You could also call it an inferrer or generator or whatever. The point is that it takes as input a sequence of preceding tokens and emits one more token to continue the sequence. danielmarkbruce 1 day ago The policy is not learned token by token during RLHF and RLVR. The reward model doesn't score token by token.
gwerbin 1 day ago Sure, you're right.But it's a policy learned from a next-token prediction task. You could also call it an inferrer or generator or whatever. The point is that it takes as input a sequence of preceding tokens and emits one more token to continue the sequence. danielmarkbruce 1 day ago The policy is not learned token by token during RLHF and RLVR. The reward model doesn't score token by token.
danielmarkbruce 1 day ago The policy is not learned token by token during RLHF and RLVR. The reward model doesn't score token by token.
Sure, you're right.
But it's a policy learned from a next-token prediction task. You could also call it an inferrer or generator or whatever. The point is that it takes as input a sequence of preceding tokens and emits one more token to continue the sequence.
The policy is not learned token by token during RLHF and RLVR. The reward model doesn't score token by token.