TypeIsPred<"input_gate_bias", NoneType>,
          TypeIsPred<"input_layer_norm_coefficients", NoneType>]>,
     Neg<Or<[
       TypeIsPred<"input_to_input_weights", NoneType>,
       TypeIsPred<"recurrent_to_input_weights", NoneType>,
       TypeIsPred<"input_gate_bias", NoneType>]>>]>>;


// TODO(b/137798843): Need to add an additional constraint for both LSTM and

      return failure();
    }

    Value filter = fc_op.getFilter();
    Value bias = fc_op.getBias();
    ElementsAttr bias_value;
    const bool is_none_bias = mlir::isa<NoneType>(bias.getType());
    if (fc_op.getFusedActivationFunction() != "NONE") return failure();

    if (!is_none_bias && !matchPattern(bias, m_Constant(&bias_value)))
      return failure();

    // Rewrite

  // on failure.
  auto build_tensor_and_buffer = [&](Value value, const int subgraph_index,
                                     const std::string& tensor_name) {
    // NoneType represents optional and may be skipped here.
    if (mlir::isa<NoneType>(value.getType())) {
      return true;
    }

    tensor_index_map.insert({value, tensors.size()});
    tensor_index_map_[subgraph_index][tensor_name] = tensors.size();

  if (getFusedActivationFunction() != "NONE") return failure();
  if (getWeightsFormat() != "DEFAULT") return failure();

  // Bias tensor is optional.
  const bool has_bias = !(!getBias() || getBias().getType().isa<NoneType>());

  // Get the tensors.
  DenseElementsAttr input_tensor, weights_tensor, bias_tensor;
  if (!matchPattern(getInput(), m_Constant(&input_tensor)) ||