Mini Activations Narrow Frontier Model Gaps
A sparse 229.9B-parameter MoE activates 9.8B parameters per token and pairs agent-native data with RL for coding, cowork, and reasoning tasks.
Jul 31, 20265 min2605.26494
3 articles on SOTA Papers
A sparse 229.9B-parameter MoE activates 9.8B parameters per token and pairs agent-native data with RL for coding, cowork, and reasoning tasks.
Progressive causal distillation and a co-designed streaming stack produce 720P rollouts at up to 16 FPS on an RTX 5090.
ABot-3DWorld 0 maps text, images, multi-view photos, and video into a shared Spatial Generative Primitive before generating panoramic exploration paths and reconstructing 3D Gaussian Splatting worlds.