Corrections to “An Efficient Two-Stage Pipelined Compute-in-Memory Macro for Accelerating Transformer Feed-Forward Networks”