return ImmutableMap.of();
    }
  }

  private static final ImmutableMap<Class<? extends Throwable>, Predicate<Throwable>> INSTANCE_OF =
      ImmutableMap.<Class<? extends Throwable>, Predicate<Throwable>>builder()
          .put(ArithmeticException.class, e -> e instanceof ArithmeticException)
          .put(
              ArrayIndexOutOfBoundsException.class,
              e -> e instanceof ArrayIndexOutOfBoundsException)

      return ImmutableMap.of();
    }
  }

  private static final ImmutableMap<Class<? extends Throwable>, Predicate<Throwable>> INSTANCE_OF =
      ImmutableMap.<Class<? extends Throwable>, Predicate<Throwable>>builder()
          .put(ArithmeticException.class, e -> e instanceof ArithmeticException)
          .put(
              ArrayIndexOutOfBoundsException.class,
              e -> e instanceof ArrayIndexOutOfBoundsException)

Java serialization is one way to implement this, however, despite its simplicity of implementation, it has several drawbacks:

- **Performance:**
Java's built-in serialization mechanism is often slower compared to other serialization solutions.
This is due to Java's use of reflection and the need to maintain a lot of metadata.

- **Size of Serialized Data:**

        }

        parameters.add(parameter);
    }

    /**
     * @return the list parameters as a Map (keyed by {@link Parameter#getName()}) that is built from
     * {@link #parameters} list on each call. In other words, the map returned is built on fly and is a copy.
     * Any change to this map is NOT reflected on list and other way around!
     */
    public Map<String, Parameter> getParameterMap() {

 * create an instance of one of the built-in implementations of {@code Graph}, use the {@link
 * GraphBuilder} class:
 *
 * <pre>{@code
 * MutableGraph<Integer> graph = GraphBuilder.undirected().build();
 * }</pre>
 *
 * <p>{@link GraphBuilder#build()} returns an instance of {@link MutableGraph}, which is a subtype

 * create an instance of one of the built-in implementations of {@code Graph}, use the {@link
 * GraphBuilder} class:
 *
 * <pre>{@code
 * MutableGraph<Integer> graph = GraphBuilder.undirected().build();
 * }</pre>
 *
 * <p>{@link GraphBuilder#build()} returns an instance of {@link MutableGraph}, which is a subtype

      return ImmutableMap.of();
    }
  }

  private static final ImmutableMap<Class<? extends Throwable>, Predicate<Throwable>> INSTANCE_OF =
      ImmutableMap.<Class<? extends Throwable>, Predicate<Throwable>>builder()
          .put(ArithmeticException.class, e -> e instanceof ArithmeticException)
          .put(
              ArrayIndexOutOfBoundsException.class,
              e -> e instanceof ArrayIndexOutOfBoundsException)

      return ImmutableMap.of();
    }
  }

  private static final ImmutableMap<Class<? extends Throwable>, Predicate<Throwable>> INSTANCE_OF =
      ImmutableMap.<Class<? extends Throwable>, Predicate<Throwable>>builder()
          .put(ArithmeticException.class, e -> e instanceof ArithmeticException)
          .put(
              ArrayIndexOutOfBoundsException.class,
              e -> e instanceof ArrayIndexOutOfBoundsException)

      return ImmutableMap.of();
    }
  }

  private static final ImmutableMap<Class<? extends Throwable>, Predicate<Throwable>> INSTANCE_OF =
      ImmutableMap.<Class<? extends Throwable>, Predicate<Throwable>>builder()
          .put(ArithmeticException.class, e -> e instanceof ArithmeticException)
          .put(
              ArrayIndexOutOfBoundsException.class,
              e -> e instanceof ArrayIndexOutOfBoundsException)

variables from your model are also going to be untrusted. That means that if
your code interacts with the filesystem, network, etc. and uses checkpointed
variables as part of those interactions (ex: using a string variable to build a
filesystem path), a maliciously created checkpoint might be able to change the
targets of those operations, which could result in arbitrary
read/write/executions.

### Running a TensorFlow server