Results
4 articlesSort
47:43ENMaking GPU Training Faster: Pipelines, Fusion and Custom KernelsCorwin · Jane StreetTwo Jane Street speakers explain how to diagnose idle GPUs, reduce memory traffic and build a parallel reduction kernel.
2:29:41ZH → English
Zhang Xiaojun Podcast· Apr 7, 2025Podcast · 4 speakersChen Jianyu: How Robot Foundation Models Can Become General-Purpose VLA SystemsGuest 陈建宇 · Host 张小珺From SayCan and RT-2 to Diffusion Policy, Tsinghua assistant professor and Epoch Robotics founder Chen Jianyu traces the path from borrowing language models to end-to-end VLA. The hard problems remain diverse data, cross-embodiment generalization, and high-frequency action control.
49:41ZH → EnglishZhang Xiaojun Podcast· Aug 17, 2024Podcast · 3 speakersFu Sheng: Large Models Are a Productivity Revolution, and Robots Have a New Window of OpportunityGuest 傅盛 · Host 张小珺Fu Sheng sees the heart of large models in their capacity for logic and reasoning. Their impact, he says, is closer to the steam engine than to another round of ordinary internet evolution. He also reflects on open models, service robots, and how entrepreneurs can face difficult periods.
27:07ENGPT-4o's Launch: Voice, Vision and a More Conversational InterfaceMatthew Berman, Mira Murati · Forward FutureMatthew Berman reviews OpenAI's May 2024 demonstrations of GPT-4o and considers what a faster multimodal assistant could enable.