/Tech1h ago

Baidu Researchers Published First Deep Learning Scaling Paper in 2017

119033.1K

Original post

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex#501inTech

I think people overestimate the importance of "frontier algorithmic secrets" obtained via espionage/SF parties/bathhouses for PRC catch-up because they don't realize that China has *never been that far behind on research*. Baidu of all people published the first scaling paper.

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex

@Object_Zero_ Use cases are too similar for AA index/Elo to become irrelevant. Intelligence is intelligence. You overestimate the value of espionage because you overestimate how far behind they were when this was all more open. Moonshot CEO did good work at Google. His mentor founded ZAI.

5:27 PM · Jul 3, 2026 · 2.3K Views

Sentiment

Sentiment building, check back later.

Cluster Engagement

Digg Deeper

No Digg Deeper questions have been answered for this story yet.

Posts from X

Most Activity

VIEWS797LIKES4

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex

and yes, notice that authors are mostly non-Chinese. Dario left Baidu just 2 years prior. This was another era. But the real catch-up was happening then, as Chinese graduates absorbed the knowledge. I doubt that US startups made much fundamental progress beyond that until ≈2022

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex

1h79740