Hybrid JIT-CUDA Graph Optimization for Low-Latency Large Language Model Inference.

Divakar Kumar Yadav, Tian Zhao

Browse the full COMPSAC paper archive.