Deep Tech

His work on reinforcement learning and embodied agents is part research, part startup, and all about learning by doing.

The V3.2 lineup marks a push for parity with Western rivals while cutting compute costs through novel architectural design.

A new class of PA-targeted drugs promises faster relief, but high prices and uneven access could slow uptake.

Affiliates ramp up takeovers and capacity buildups in line with China’s national ambitions.

The company’s push into multimodal data collection and reinforcement learning signals its aim to control the full robotics stack.

LingGuang goes beyond text-only chat, generating visuals, charts, 3D models, and even mini apps from natural language commands.

Among the updates is the RDK S600, a platform built to support edge deployment of large models on robots.

A learning paradigm rooted in human motion capture could offer a stronger path forward.