REASONING COMPILER: LLM-Guided Optimizations for Efficient Model Serving

Annabelle Sujun Tang, Christopher Priebe, Rohan Mahapatra +2

While model serving has unlocked unprecedented capabilities, the high cost of serving large-scale models continues to be a significant barrier to widespread accessibility and rapid…