Submitted by Quantong Qiu 9 Flux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference Long-Context Model Laboratory 6 2