#KimiK3 These past couple of days, people have been discussing Chen Guangyu, the 17-year-old prodigy who used to work at Kimi. Yesterday, he officially announced that he was leaving Kimi. A year ago, Moonshot AI brought him onto its core team, and he was a co-first author of the paper “Attention Residuals.” The paper proposed an improved approach to residual connections and even earned public praise from Elon Musk. It also provides important theoretical support for the Kimi K3 model.
What’s even more surprising is that he only started learning about machine learning in 2024. Before long, he had gone from just learning about Transformers to becoming an author of papers on cutting-edge architectures.
Judging by what he wrote, his departure from Kimi seems to have been amicable and on good terms. He did say he’s heading to Silicon Valley, though—it’s unclear whether he’s going there to study or start a company 🤔🤔
What’s even more surprising is that he only started learning about machine learning in 2024. Before long, he had gone from just learning about Transformers to becoming an author of papers on cutting-edge architectures.
Judging by what he wrote, his departure from Kimi seems to have been amicable and on good terms. He did say he’s heading to Silicon Valley, though—it’s unclear whether he’s going there to study or start a company 🤔🤔