Can't use SpectralNormalization wrapper layer.

[willgryan]

Using SpectralNormalization in a sequential Keras model results in this error:

2021-05-17 20:47:45.630609: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_1 with frame_id = 0 and iter_id = 0 with error: Internal: CreateMLCFullyConnectedLayer: Failed to create MLCFullyConnectedLayer for n20 = sequential/spectral_normalization/dense/MLCMatMul_1 (MLCMatMul) (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:45.697962: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/MLCSubgraphOp_0_6 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sub_2 (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:45.698496: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/sequential/MLCSubgraphOp_0_9 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sequential/spectral_normalization_3/dense_3/BiasAdd_1 (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:45.797123: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_10 with frame_id = 0 and iter_id = 0 with error: Internal: GetTensorParam: Not found MLC weight tensor in the given layer. (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:45.867406: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_2 with frame_id = 0 and iter_id = 0 with error: Internal: CreateMLCFullyConnectedLayer: Failed to create MLCFullyConnectedLayer for n40 = sequential/spectral_normalization/dense/MLCMatMul_2 (MLCMatMul) (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.009352: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_11 with frame_id = 0 and iter_id = 0 with error: Internal: CreateMLCFullyConnectedLayer: Failed to create MLCFullyConnectedLayer for n15 = sequential/spectral_normalization/dense/MLCMatMul_3 (MLCMatMul) (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.068199: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/MLCSubgraphOp_0_15 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: add_2 (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.090369: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_17 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sequential/spectral_normalization_3/dense_3/MLCMatMul_2 (error will be reported 5 times unless TF_MLC_LOGGING=1).
1/222 [..............................] - ETA: 5:58 - d_loss: 2.7717 - g_loss: 1.92832021-05-17 20:47:46.342171: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_0 with frame_id = 0 and iter_id = 0 with error: Internal: CreateMLCFullyConnectedLayer: Failed to create MLCFullyConnectedLayer for n37 = sequential_1/dense_5/MLCMatMul (MLCMatMul) (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.468152: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_1 with frame_id = 0 and iter_id = 0 with error: Internal: CreateMLCFullyConnectedLayer: Failed to create MLCFullyConnectedLayer for n20 = sequential/spectral_normalization/dense/MLCMatMul_1 (MLCMatMul) (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.524743: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/MLCSubgraphOp_0_6 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sub_2 (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.524987: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/sequential/MLCSubgraphOp_0_9 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sequential/spectral_normalization_3/dense_3/BiasAdd_1 (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.525164: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/sequential/MLCSubgraphOp_0_8 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sequential/spectral_normalization_3/dense_3/BiasAdd (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.668207: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_10 with frame_id = 0 and iter_id = 0 with error: Internal: GetTensorParam: Not found MLC weight tensor in the given layer. (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.719739: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_2 with frame_id = 0 and iter_id = 0 with error: Internal: CreateMLCFullyConnectedLayer: Failed to create MLCFullyConnectedLayer for n40 = sequential/spectral_normalization/dense/MLCMatMul_2 (MLCMatMul) (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.847721: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_11 with frame_id = 0 and iter_id = 0 with error: Internal: CreateMLCFullyConnectedLayer: Failed to create MLCFullyConnectedLayer for n15 = sequential/spectral_normalization/dense/MLCMatMul_3 (MLCMatMul) (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.910711: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/MLCSubgraphOp_0_15 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: add_2 (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:46.910992: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_17 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sequential/spectral_normalization_3/dense_3/MLCMatMul_2 (error will be reported 5 times unless TF_MLC_LOGGING=1).
2/222 [..............................] - ETA: 3:00 - d_loss: 2.5948 - g_loss: 5.41092021-05-17 20:47:47.160554: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_0 with frame_id = 0 and iter_id = 0 with error: Internal: CreateMLCFullyConnectedLayer: Failed to create MLCFullyConnectedLayer for n37 = sequential_1/dense_5/MLCMatMul (MLCMatMul) (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:47.286522: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_1 with frame_id = 0 and iter_id = 0 with error: Internal: CreateMLCFullyConnectedLayer: Failed to create MLCFullyConnectedLayer for n20 = sequential/spectral_normalization/dense/MLCMatMul_1 (MLCMatMul) (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:47.348304: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/MLCSubgraphOp_0_6 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sub_2 (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:47.348576: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/sequential/MLCSubgraphOp_0_8 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sequential/spectral_normalization_3/dense_3/BiasAdd (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:47.348580: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph gradient_tape/sequential/MLCSubgraphOp_0_9 with frame_id = 0 and iter_id = 0 with error: Internal: PerformGradientPassNodeRoutine: Failed to find forward-pass output for node: sequential/spectral_normalization_3/dense_3/BiasAdd_1 (error will be reported 5 times unless TF_MLC_LOGGING=1).
2021-05-17 20:47:47.490132: I tensorflow/compiler/tf2mlcompute/kernels/mlc_subgraph_op.cc:326] Compute: Failed in processing TensorFlow graph MLCSubgraphOp_0_10 with