El artículo de DeepSeek-R1 aparece en la portada de Nature, impulsando el proceso de transparencia en IA.

【DeepSeek-R1 paper featured on the cover of Nature, advancing AI transparency】The DeepSeek-R1 paper has been published as a cover article in Nature, with DeepSeek founder and CEO Liang Wenfeng as the corresponding author. The research team demonstrated through experiments that the reasoning capabilities of large language models can be enhanced through pure reinforcement learning, reducing the workload of human input, and outperforming models trained with traditional methods in tasks such as mathematics and programming. DeepSeek-R1 has received 91.1k stars on GitHub, earning praise from developers worldwide. An assistant professor at Carnegie Mellon University commented on its evolution from a powerful but opaque solutions seeker to a system capable of human-like dialogue. Nature acknowledged it in an editorial as the first mainstream LLM published after peer review, marking a promising step towards transparency; peer review helps clarify how LLMs work, assess their effectiveness, and enhance model safety.