@zhengwu777
researcher @ Meta TBD, PhD @ UC Berkeley
AI
Kun Chen tested non-mainstream models and highlighted Muse Spark 1.3 performance.