Bluesky Thread

DeepSeek published about R1 in Nature

September 17, 2025 View original thread

DeepSeek published about R1 in Nature

www.nature.com/articles/s41...

but the supplementary information is far more detailed. it includes, lots of details on everything from datasets, data processing, hyperparameters, and more

static-content.springer.com/esm/art%3A10...

DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning - Nature

A new artificial intelligence model, DeepSeek-R1, is introduced, demonstrating that the reasoning abilities of large language models can be incentivized through pure reinforcement learning, removing t...

40 7

obligatory bsky.app/profile/timk...

Tim Kellogg @timkellogg.me

during COVID i started reading a ton of papers, especially around COVID

Nature was an exceedingly good outlet, a tremendous noise reducer in an otherwise noisy space

but for AI, they miss on repeat. it’s almost impressive how bad they are

1

More like this

DeepSeek is shipping a theorem prover (automate math proofs)

DeepSeek did the “one more thing” 🙄

RLP: Reinforcement Learning in Pre-Training