Machine Learning25 min reading time

How a Frontier Model Gets Built, Read from the Kimi K3 Report

Towards Data Science
Read the full article
Moonshot AI released a detailed technical report on their Kimi K3 model, a 2.8-trillion-parameter mixture-of-experts transformer with innovations enabling million-token context windows and improved scaling efficiency. The report reveals how architectural tweaks and training refinements reduced compute costs while maintaining performance near top closed models.

More on this story


More in Machine Learning

Machine Learning5 min read

Researchers from Stanford and UC Berkeley Introduce TokenGate: Token-Gated Attention Makes Long-Context Prefill Up to 11.3x Faster

MarkTechPost
Machine Learning2 min read

Jack Dorsey’s Block made an AI model that talks to you like an awkward new friend

Covered by 5 sources
Machine Learning6 min read

AI21 Labs Releases LIFT-Bench to Curb Reward Hacking in RL Training

Unite.AI