}

  /**
   * Returns {@code x}, rounded to a {@code double} with the specified rounding mode. If {@code x}
   * is precisely representable as a {@code double}, its {@code double} value will be returned;
   * otherwise, the rounding will choose between the two nearest representable values with {@code
   * mode}.
   *
   * <p>For the case of {@link RoundingMode#HALF_DOWN}, {@code HALF_UP}, and {@code HALF_EVEN},

   * is precisely representable as a {@code double}, its {@code double} value will be returned;
   * otherwise, the rounding will choose between the two nearest representable values with {@code
   * mode}.
   *
   * <p>For the case of {@link RoundingMode#HALF_EVEN}, this implementation uses the IEEE 754
   * default rounding mode: if the two nearest representable values are equally near, the one with

   * is precisely representable as a {@code double}, its {@code double} value will be returned;
   * otherwise, the rounding will choose between the two nearest representable values with {@code
   * mode}.
   *
   * <p>For the case of {@link RoundingMode#HALF_EVEN}, this implementation uses the IEEE 754
   * default rounding mode: if the two nearest representable values are equally near, the one with

</li>
<li>
<code>x</code> is an untyped <a href="#Constants">constant</a>
<a href="#Representability">representable</a>
by a value of type <code>T</code>.
</li>
</ul>


<h3 id="Representability">Representability</h3>

<p>
A <a href="#Constants">constant</a> <code>x</code> is <i>representable</i>
by a value of type <code>T</code> if one of the following conditions applies:
</p>

<ul>

<a href="#Complex_numbers">components</a> <code>real(x)</code> and <code>imag(x)</code>
are representable by values of <code>T</code>'s component type (<code>float32</code> or
<code>float64</code>).
</li>
</ul>

<p>
If <code>T</code> is a type parameter,
<code>x</code> is representable by a value of type <code>T</code> if <code>x</code> is representable
by a value of each type in <code>T</code>'s type set.
</p>

<pre>

   * charset} parameter will be present on the new instance regardless of the number set on this
   * one.
   *
   * <p>If a charset must be specified that is not supported on this JVM (and thus is not
   * representable as a {@link Charset} instance), use {@link #withParameter}.
   */
  public MediaType withCharset(Charset charset) {
    checkNotNull(charset);
    MediaType withCharset = withParameter(CHARSET_ATTRIBUTE, charset.name());

        `tf.GradientTape` inside a `tf.function`.
    *   Changed the default step size in `gradient_checker_v2.compute_gradients`
        to be exactly representable as a binary floating point numbers. This
        avoids poluting gradient approximations needlessly, which is some cases
        leads to false negatives in op gradient tests.