AURA: Always-On Understanding and Real-Time Assistance via Video Streams
Paper • 2604.04184 • Published • 43
Xinyu Fu
cynricfu
·
AI & ML interests
None yet
Recent Activity
upvoted a paper 5 days ago
AURA: Always-On Understanding and Real-Time Assistance via Video Streams upvoted a paper 6 months ago
MathCanvas: Intrinsic Visual Chain-of-Thought for Multimodal
Mathematical Reasoning commentedon an article 8 months ago
Accelerate ND-Parallel: A guide to Efficient Multi-GPU TrainingOrganizations
None yet
upvoted a paper 5 days ago
upvoted a paper 6 months ago
commentedon Accelerate ND-Parallel: A guide to Efficient Multi-GPU Training 8 months ago
May I ask why is dp_world_size realted to tp_size?
Based on my understanding, it should be just dp_world_size = dp_shard_size * dp_replicate_size. Or am I missing some points here?
liked a
Space about 1 year ago