Reaction
Distillation's role in Chinese AI model progress
One commenter says it's hard to separate distillation's contribution from gains made through internal reinforcement learning.
TLDR
One commenter says they can't cleanly determine how much Chinese model progress comes from distillation versus increased internal reinforcement learning. In a reply, another commenter argues that distillation should be legal but says they don't want to encounter Claude everywhere.
Combined views
2K
3 Sources, first seen ago
