๐ Excited to see everyoneโs enthusiasm for deploying DeepSeek-R1! Here are our recommended settings for the best experience:
โข No system prompt
โข Temperature: 0.6
โข Official prompts for search & file upload: https://t.co/I5CqmSzkTQ
โข Guidelines to mitigate model bypass thinking: https://t.co/sAXK5U6OEr
The official DeepSeek deployment runs the same model as the open-source versionโenjoy the full DeepSeek-R1 experience! ๐
Absolutely mind-numbing contradictions in this article by Dario Amodei, the CEO of Anthropic (the company behind Claude AI).
He simultaneously argues that Deepseek doesn't matter much, yet begs the U.S. government to ramp up measures to prevent Chinese AI development because it's "existentially important".
And he explains it all as a defense of "democracy vs authoritarianism" yet advocates for the most authoritarian vision of AI you can possibly imagine: using state power to crush open AI development and force global dependence on closed U.S. AI models like his.
Which all goes to show just how threatened U.S. AI companies feel - when you're begging for government intervention to stop others from developing AI, you're effectively admitting defeat since it shows you don't think you can compete on your own merits.