const int input_idx = 0;
  EXPECT_THAT(subgraph->tensors[input_idx]->type, Eq(TensorType_INT8));
  EXPECT_THAT(subgraph->tensors[input_idx]->name, Eq("input"));
  EXPECT_THAT(subgraph->tensors[input_idx]->quantization->scale, SizeIs(1));
  EXPECT_THAT(subgraph->tensors[input_idx]->quantization->zero_point,
              SizeIs(1));
  EXPECT_THAT(subgraph->tensors[input_idx]->quantization->scale[0],

      indices.getType().getRank() != 1 || port[1] >= indices.getNumElements()) {
    return {};
  }

  int64_t input_idx = indices.getValues<IntegerAttr>()[port[1]].getInt();
  if (input_idx >= params_ty.getDimSize(0)) return {};

  ValuePort req(params);
  req.port.push_back(input_idx);
  return req;
}

ValuePort ComputeInputComponentFor(Operation* op, ArrayRef<unsigned int> port) {

*   [0, 255] for unsigned, num_bits = 8.

If range_given == False, the initial input_min, input_max will be determined
automatically as the minimum and maximum values in the input tensor, otherwise
the specified values of input_min, input_max are used.

Note: If the input_min, input_max are specified, they do not need to equal the
actual minimum and maximum values in the tensor. e.g. in some cases it may be

        int64_t size =
            padding_low[i] + shape[i] + (shape[i] - 1) * padding_interm[i];
        padding_high.push_back(input_dim - size);
      } else {
        dims_to_reverse.push_back(i);
        padding_high.push_back(input_dim - begin_indices[i] - 1);
        padding_interm.push_back(-strides[i] - 1);

        // Pad the lower dimension up to the expected input shape.