Long-Context Modeling with Dynamic Hierarchical Sparse Attention for On-Device LLMs

Re-assign community