Binkg(@dummerspast39):Kimi K3.1 Pro weights are out. New: They replaced all 24 Gated MLA layers from K3 with Gated DeepSeek Sparse Attention. The other 69 KDA layers remain unchanged. DSA uses a top-k of just 2,048 for a model with 1M context. DeepSeek is a year ahead when it comes to architecture.

Binkg

@dummerspast39

anti ai lobby china shill big hater

가입 August 2024

284 팔로잉 중 18.4K 팬

Binkg@dummerspast39

2026.11.04 15:33

Kimi K3.1 Pro weights are out. New: They replaced all 24 Gated MLA layers from K3 with Gated DeepSeek Sparse Attention. The other 69 KDA layers remain unchanged. DSA uses a top-k of just 2,048 for a model with 1M context. DeepSeek is a year ahead when it comes to architecture.