Why torch nn.Linear can't convert to Pb file?

[lovekittynine]

when the inputs of nn.Linear is three-dimension （batchxseq_lenxdim_size）like the "FeedForward" of Transformer，it occurs ValueError: Shape must be rank 2 but is rank 3 for 'MatMul_2' (op: 'MatMul') with input shapes: [1,20,300], [300,300]. How to Solve it? Thanks very much!