Different batch size results to different result for gptq_marlin_gemm kernel.

Hi, i notice a problem. lets say for same query, when batch size is 2, it will result to different calculate output compare to batch size is 1. Thought the different is minor, so i want to know if there is any operation can cause this problem? i am really apperate for the help!!