fix(cuda): accept fp16/bf16 router_probs, scales and biases in QMoE - #1602

fix(cuda): accept fp16/bf16 router_probs, scales and biases in QMoE#1602

justinchuby merged 1 commit into

mainjustinchuby/onnx-genai:mainfrom

fix/qmoe-fp16-router-probsjustinchuby/onnx-genai:fix/qmoe-fp16-router-probsCopy head branch name to clipboard