Fetching the paper…

Tree Attention: Topology-aware Decoding for Long-Context Attention on GPU clusters · Around