Tag
MOSS-VL-Realtime is a realtime streaming vision-language model that processes continuous video frames, supports interruptible interaction, proactive silence, and dynamic correction, with timestamp-aware encoding and a 256K context window.
Call for papers for the MOSS workshop at COLM focusing on small-scale research, with a deadline of June 30.
Garry Tan highlights that retrieval is the key bottleneck in voice AI and introduces Moss, an open-source tool achieving sub-10ms vector search, alongside a hackathon at YC office on June 6-7.