return absl::OkStatus();
        case 16:
          *tf_dtype = TF_DataType::TF_INT16;
          return absl::OkStatus();
        case 32:
          *tf_dtype = TF_DataType::TF_INT32;
          return absl::OkStatus();
        case 64:
          *tf_dtype = TF_DataType::TF_INT64;
          return absl::OkStatus();
        default:
          return tensorflow::errors::InvalidArgument("Unsupported Int bits: ",

    return absl::OkStatus();
  }
  Status SetAttrFloat(const char* attr_name, float value) override {
    op_->node_builder.Attr(attr_name, value);
    return absl::OkStatus();
  }
  Status SetAttrBool(const char* attr_name, bool value) override {
    op_->node_builder.Attr(attr_name, value);
    return absl::OkStatus();
  }

        "Trying to register filesystem without `cleanup` operation");

  return OkStatus();
}

// Validates the random access file operations supplied by the plugin.
static Status ValidateHelper(const TF_RandomAccessFileOps* ops) {
  if (ops == nullptr) {
    // We allow filesystems where files can only be written to (from TF code)
    return OkStatus();
  }

  if (ops->cleanup == nullptr)

  return IsRefType(dt)
             ? InvalidArgument("Output ", idx, " of node '", node->name(),
                               "' has a reference type ", DataTypeString(dt))
             : absl::OkStatus();
}

// Converts `ninputs` and `inputs` into `inputs_tensors` and `input_nodes` and
// does various checks while doing so. `input_nodes` will contain the same

  forward_op_->inputs.push_back(input);
  return absl::OkStatus();
}
Status AddInputList(AbstractOperation* op_,
                    absl::Span<AbstractTensorHandle* const> inputs,
                    ForwardOperation* forward_op_) {
  TF_RETURN_IF_ERROR(op_->AddInputList(inputs));
  for (auto input : inputs) {
    forward_op_->inputs.push_back(input);
  }
  return absl::OkStatus();
}

    DCHECK(grad_outputs[0]);
    grad_inputs[0] = grad_outputs[0];
    grad_inputs[1] = grad_outputs[0];

    grad_inputs[0]->Ref();
    grad_inputs[1]->Ref();
    return absl::OkStatus();
  }
  ~AddGradientFunction() override {}
};

class ExpGradientFunction : public GradientFunction {
 public:
  explicit ExpGradientFunction(AbstractTensorHandle* exp) : exp_(exp) {
    exp->Ref();
  }

  VALIDATE_MEMBER(TP_OptimizerRegistrationParams, params, device_type);
  return absl::OkStatus();
}

absl::Status ValidateTPOptimizer(const TP_Optimizer& optimizer) {
  VALIDATE_STRUCT_SIZE(TP_Optimizer, optimizer, TP_OPTIMIZER_STRUCT_SIZE);
  VALIDATE_MEMBER(TP_Optimizer, optimizer, optimize_func);
  return absl::OkStatus();
}

absl::Status ValidateTPOptimizerConfigs(const TP_OptimizerConfigs& configs) {

        void* opaque_plugin_resource = create_func(create_func_args);
        *new_resource = new tensorflow::PluginResource(
            opaque_plugin_resource, plugin_resource_name, delete_func);
        return tensorflow::OkStatus();
      });

  if (cc_status.ok()) {
    tf_plugin_resource_ptr.reset(tf_plugin_resource);
    *result_plugin_resource = tf_plugin_resource_ptr->GetOpaquePluginResource();
  } else {

        op_type, output_tensors, input_tensor_id, input_dtypes,
        backward_function_getter, backward_function_deleter);
    return absl::OkStatus();
  }
  if (!ShouldRecord(input_tensor_id, input_dtypes)) {
    return absl::OkStatus();
  }

  // We may need to allocate zero inputs for trainable dtypes we don't have JVPs
  // for. Make sure they get cleaned up.

  if (read > 0) *result = StringPiece(scratch, read);
  return StatusFromTF_Status(plugin_status.get());
}

Status ModularRandomAccessFile::Name(StringPiece* result) const {
  *result = filename_;
  return OkStatus();
}

Status ModularWritableFile::Append(StringPiece data) {
  if (ops_->append == nullptr)
    return errors::Unimplemented(tensorflow::strings::StrCat(
        "Append() not implemented for ", filename_));