"FhaLiFa team on AMD AI League Hackathon"
Team Idea
An agent that autotunes vLLM serving configs on AMD MI300X GPUs — instead of brute-forcing every quantization and batch-size combination, an LLM planner picks the next config to test based on what's still uncertain, then recommends the fastest config that doesn't sacrifice output quality
Team Members
Sakibul Huda Onnay
FhaLiFa
Submission
No submission yet