Practice
Learn
← Quit
0 / 0 correct
LLMs & AI
Hard
1 of 20
What is DPO (Direct Preference Optimization)'s main advantage over PPO-based RLHF?
A
It trains the tokenizer
B
It trains directly on preference pairs with a simple classification-style loss, with no separate reward model or RL loop
C
It needs no preference data
D
It's only for images