3 ms·
Do we know if other models have started detecting and poisoning training/fine tuning that these Chinese models seem to use for alignment, I’d certainly be doing
by andy_ppp 6mo ago
Do we know if other models have started detecting and poisoning training/fine tuning that these Chinese models seem to use for alignment, I’d certainly be doing some naughty stuff to keep my moat if I was Anthropic or OpenAI…
- storus 6mo agoThey no longer show reasoning traces and are throttling more aggressively.