artifacts { compile jar1, jar2 }
            }

            abstract class Hasher implements TransformAction<TransformParameters.None> {
                private int count

                Hasher() {
                    println "Creating Transform"
                }

                @InputArtifact

                    println "Transformed \$input.name to \$output.name into \$outputDirectory"
                    output.text = String.valueOf(input.length())
                }
            }
            abstract class Hasher implements TransformAction<Parameters> {
                interface Parameters extends TransformParameters {
                    @Input
                    Property<String> getTarget()
                }

    exclude_unset: bool = False,
    exclude_defaults: bool = False,
    exclude_none: bool = False,
    is_coroutine: bool = True,
) -> Any:
    if field:
        errors = []
        if not hasattr(field, "serialize"):
            # pydantic v1
            response_content = _prepare_response_content(
                response_content,
                exclude_unset=exclude_unset,

  return changed;
}

bool ShapeInference::InferShapeForFunctionAttachedToXlaHostCompute(
    XlaHostComputeOp op) {
  const std::string kShapeInferenceGraph = "shape_inference_graph";
  if (!op->hasAttr(kShapeInferenceGraph)) {
    return false;
  }

  ModuleOp module = op->getParentOfType<ModuleOp>();
  func::FuncOp func = module.lookupSymbol<func::FuncOp>(

  // attribute. This provides a hook for the user to represent the variable
  // tensor in the MLIR level.
  if (auto* inst = value.getDefiningOp();
      inst && inst->hasAttr("tfl.is_variable")) {
    is_variable = true;
  }

  bool has_rank = type.hasRank();

  if (shape_signature.empty()) {
    return tflite::CreateTensor(