Python tensorflow.contrib.slim.one_hot_encoding() Examples

def test_create_summaries_is_runnable(self):
    ocr_model = self.create_model()
    data = data_provider.InputEndpoints(
        images=self.fake_images,
        images_orig=self.fake_images,
        labels=self.fake_labels,
        labels_one_hot=slim.one_hot_encoding(self.fake_labels,
                                             self.num_char_classes))
    endpoints = ocr_model.create_base(
        images=self.fake_images, labels_one_hot=None)
    charset = create_fake_charset(self.num_char_classes)
    summaries = ocr_model.create_summaries(
        data, endpoints, charset, is_training=False)
    with self.test_session() as sess:
      sess.run(tf.global_variables_initializer())
      sess.run(tf.local_variables_initializer())
      tf.tables_initializer().run()
      sess.run(summaries)  # just check it is runnable 

def test_create_summaries_is_runnable(self):
    ocr_model = self.create_model()
    data = data_provider.InputEndpoints(
        images=self.fake_images,
        images_orig=self.fake_images,
        labels=self.fake_labels,
        labels_one_hot=slim.one_hot_encoding(self.fake_labels,
                                             self.num_char_classes))
    endpoints = ocr_model.create_base(
        images=self.fake_images, labels_one_hot=None)
    charset = create_fake_charset(self.num_char_classes)
    summaries = ocr_model.create_summaries(
        data, endpoints, charset, is_training=False)
    with self.test_session() as sess:
      sess.run(tf.global_variables_initializer())
      sess.run(tf.local_variables_initializer())
      tf.tables_initializer().run()
      sess.run(summaries)  # just check it is runnable 

def char_predictions(self, chars_logit):
    """Returns confidence scores (softmax values) for predicted characters.

    Args:
      chars_logit: chars logits, a tensor with shape
        [batch_size x seq_length x num_char_classes]

    Returns:
      A tuple (ids, log_prob, scores), where:
        ids - predicted characters, a int32 tensor with shape
          [batch_size x seq_length];
        log_prob - a log probability of all characters, a float tensor with
          shape [batch_size, seq_length, num_char_classes];
        scores - corresponding confidence scores for characters, a float
        tensor
          with shape [batch_size x seq_length].
    """
    log_prob = utils.logits_to_log_prob(chars_logit)
    ids = tf.to_int32(tf.argmax(log_prob, dimension=2), name='predicted_chars')
    mask = tf.cast(
        slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
    all_scores = tf.nn.softmax(chars_logit)
    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
    return ids, log_prob, scores 

def char_predictions(self, chars_logit):
    """Returns confidence scores (softmax values) for predicted characters.

    Args:
      chars_logit: chars logits, a tensor with shape
        [batch_size x seq_length x num_char_classes]

    Returns:
      A tuple (ids, log_prob, scores), where:
        ids - predicted characters, a int32 tensor with shape
          [batch_size x seq_length];
        log_prob - a log probability of all characters, a float tensor with
          shape [batch_size, seq_length, num_char_classes];
        scores - corresponding confidence scores for characters, a float
        tensor
          with shape [batch_size x seq_length].
    """
    log_prob = utils.logits_to_log_prob(chars_logit)
    ids = tf.to_int32(tf.argmax(log_prob, axis=2), name='predicted_chars')
    mask = tf.cast(
      slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
    all_scores = tf.nn.softmax(chars_logit)
    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
    return ids, log_prob, scores 

def test_create_summaries_is_runnable(self):
    ocr_model = self.create_model()
    data = data_provider.InputEndpoints(
        images=self.fake_images,
        images_orig=self.fake_images,
        labels=self.fake_labels,
        labels_one_hot=slim.one_hot_encoding(self.fake_labels,
                                             self.num_char_classes))
    endpoints = ocr_model.create_base(
        images=self.fake_images, labels_one_hot=None)
    charset = create_fake_charset(self.num_char_classes)
    summaries = ocr_model.create_summaries(
        data, endpoints, charset, is_training=False)
    with self.test_session() as sess:
      sess.run(tf.global_variables_initializer())
      sess.run(tf.local_variables_initializer())
      tf.tables_initializer().run()
      sess.run(summaries)  # just check it is runnable 

def class_and_spatial_loss(logits, onehot_labels, weights, weights2):
    logits_shape = tf.shape(logits)
    onehot_labels_shape = tf.shape(onehot_labels)
    image_labels = tf.reshape(onehot_labels, logits_shape)
    class_loss = tf.losses.softmax_cross_entropy(
        onehot_labels=onehot_labels,
        logits=tf.reshape(logits, [-1, onehot_labels_shape[-1]]),
        weights=weights * weights2
    )

    image_weights = tf.reshape(weights, [logits_shape[0], logits_shape[1], logits_shape[2], 1])
    predict_class = tf.argmax(logits, axis=3)
    predict_class = slim.one_hot_encoding(predict_class, onehot_labels_shape[-1], 1.0, 0.0)
    union = to_float(to_bool(predict_class + image_labels)) * image_weights
    intersection = to_float(tf.logical_and(to_bool(predict_class), to_bool(image_labels))) * image_weights
    label_on = to_float(tf.greater(tf.reduce_sum(image_labels, axis=[1, 2]), 0))
    spatial_loss = ((tf.reduce_sum(intersection, axis=[1, 2]) + 1) / (tf.reduce_sum(union, axis=[1, 2]) + 1))
    spatial_loss = tf.reduce_mean(-tf.log(spatial_loss) * label_on)

    return class_loss + spatial_loss 

def char_predictions(self, chars_logit):
    """Returns confidence scores (softmax values) for predicted characters.

    Args:
      chars_logit: chars logits, a tensor with shape
        [batch_size x seq_length x num_char_classes]

    Returns:
      A tuple (ids, log_prob, scores), where:
        ids - predicted characters, a int32 tensor with shape
          [batch_size x seq_length];
        log_prob - a log probability of all characters, a float tensor with
          shape [batch_size, seq_length, num_char_classes];
        scores - corresponding confidence scores for characters, a float
        tensor
          with shape [batch_size x seq_length].
    """
    log_prob = utils.logits_to_log_prob(chars_logit)
    ids = tf.to_int32(tf.argmax(log_prob, axis=2), name='predicted_chars')
    mask = tf.cast(
      slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
    all_scores = tf.nn.softmax(chars_logit)
    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
    return ids, log_prob, scores 

def char_predictions(self, chars_logit):
    """Returns confidence scores (softmax values) for predicted characters.

    Args:
      chars_logit: chars logits, a tensor with shape
        [batch_size x seq_length x num_char_classes]

    Returns:
      A tuple (ids, log_prob, scores), where:
        ids - predicted characters, a int32 tensor with shape
          [batch_size x seq_length];
        log_prob - a log probability of all characters, a float tensor with
          shape [batch_size, seq_length, num_char_classes];
        scores - corresponding confidence scores for characters, a float
        tensor
          with shape [batch_size x seq_length].
    """
    log_prob = utils.logits_to_log_prob(chars_logit)
    ids = tf.to_int32(tf.argmax(log_prob, axis=2), name='predicted_chars')
    mask = tf.cast(
      slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
    all_scores = tf.nn.softmax(chars_logit)
    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
    return ids, log_prob, scores 

def test_create_summaries_is_runnable(self):
    ocr_model = self.create_model()
    data = data_provider.InputEndpoints(
        images=self.fake_images,
        images_orig=self.fake_images,
        labels=self.fake_labels,
        labels_one_hot=slim.one_hot_encoding(self.fake_labels,
                                             self.num_char_classes))
    endpoints = ocr_model.create_base(
        images=self.fake_images, labels_one_hot=None)
    charset = create_fake_charset(self.num_char_classes)
    summaries = ocr_model.create_summaries(
        data, endpoints, charset, is_training=False)
    with self.test_session() as sess:
      sess.run(tf.global_variables_initializer())
      sess.run(tf.local_variables_initializer())
      tf.tables_initializer().run()
      sess.run(summaries)  # just check it is runnable 

def test_create_summaries_is_runnable(self):
    ocr_model = self.create_model()
    data = data_provider.InputEndpoints(
        images=self.fake_images,
        images_orig=self.fake_images,
        labels=self.fake_labels,
        labels_one_hot=slim.one_hot_encoding(self.fake_labels,
                                             self.num_char_classes))
    endpoints = ocr_model.create_base(
        images=self.fake_images, labels_one_hot=None)
    charset = create_fake_charset(self.num_char_classes)
    summaries = ocr_model.create_summaries(
        data, endpoints, charset, is_training=False)
    with self.test_session() as sess:
      sess.run(tf.global_variables_initializer())
      sess.run(tf.local_variables_initializer())
      tf.tables_initializer().run()
      sess.run(summaries)  # just check it is runnable 

def char_predictions(self, chars_logit):
    """Returns confidence scores (softmax values) for predicted characters.

    Args:
      chars_logit: chars logits, a tensor with shape
        [batch_size x seq_length x num_char_classes]

    Returns:
      A tuple (ids, log_prob, scores), where:
        ids - predicted characters, a int32 tensor with shape
          [batch_size x seq_length];
        log_prob - a log probability of all characters, a float tensor with
          shape [batch_size, seq_length, num_char_classes];
        scores - corresponding confidence scores for characters, a float
        tensor
          with shape [batch_size x seq_length].
    """
    log_prob = utils.logits_to_log_prob(chars_logit)
    ids = tf.to_int32(tf.argmax(log_prob, axis=2), name='predicted_chars')
    mask = tf.cast(
      slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
    all_scores = tf.nn.softmax(chars_logit)
    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
    return ids, log_prob, scores 

def test_create_summaries_is_runnable(self):
    ocr_model = self.create_model()
    data = data_provider.InputEndpoints(
        images=self.fake_images,
        images_orig=self.fake_images,
        labels=self.fake_labels,
        labels_one_hot=slim.one_hot_encoding(self.fake_labels,
                                             self.num_char_classes))
    endpoints = ocr_model.create_base(
        images=self.fake_images, labels_one_hot=None)
    charset = create_fake_charset(self.num_char_classes)
    summaries = ocr_model.create_summaries(
        data, endpoints, charset, is_training=False)
    with self.test_session() as sess:
      sess.run(tf.global_variables_initializer())
      sess.run(tf.local_variables_initializer())
      tf.tables_initializer().run()
      sess.run(summaries)  # just check it is runnable 

def _top(self, prediction, truth, num_tops=1):
        # a full sort using top_k
        values, indices = tf.nn.top_k(prediction, self.num_classes)
        # cut-off threshold
        thresholds = values[:, (num_tops - 1):num_tops]
        # if > threshold, weight = 1, else weight = 0
        valids = tf.cast(prediction > thresholds, tf.float32)
        # ties should have weight = 1 / num_ties
        ties = tf.equal(prediction, thresholds)
        num_ties = tf.reduce_sum(
            tf.cast(ties, tf.float32), axis=-1, keepdims=True)
        num_ties = tf.py_func(
            self._warn_ties, [ties, num_ties, thresholds],
            tf.float32, stateful=False)
        num_ties = tf.tile(num_ties, [1, self.num_classes])
        weights = tf.where(ties, 1 / num_ties, valids)
        return slim.one_hot_encoding(truth, self.num_classes) * weights 

def char_predictions(self, chars_logit):
    """Returns confidence scores (softmax values) for predicted characters.

    Args:
      chars_logit: chars logits, a tensor with shape
        [batch_size x seq_length x num_char_classes]

    Returns:
      A tuple (ids, log_prob, scores), where:
        ids - predicted characters, a int32 tensor with shape
          [batch_size x seq_length];
        log_prob - a log probability of all characters, a float tensor with
          shape [batch_size, seq_length, num_char_classes];
        scores - corresponding confidence scores for characters, a float
        tensor
          with shape [batch_size x seq_length].
    """
    log_prob = utils.logits_to_log_prob(chars_logit)
    ids = tf.to_int32(tf.argmax(log_prob, axis=2), name='predicted_chars')
    mask = tf.cast(
      slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
    all_scores = tf.nn.softmax(chars_logit)
    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
    return ids, log_prob, scores 

def char_predictions(self, chars_logit):
    """Returns confidence scores (softmax values) for predicted characters.

    Args:
      chars_logit: chars logits, a tensor with shape
        [batch_size x seq_length x num_char_classes]

    Returns:
      A tuple (ids, log_prob, scores), where:
        ids - predicted characters, a int32 tensor with shape
          [batch_size x seq_length];
        log_prob - a log probability of all characters, a float tensor with
          shape [batch_size, seq_length, num_char_classes];
        scores - corresponding confidence scores for characters, a float
        tensor
          with shape [batch_size x seq_length].
    """
    log_prob = utils.logits_to_log_prob(chars_logit)
    ids = tf.to_int32(tf.argmax(log_prob, axis=2), name='predicted_chars')
    mask = tf.cast(
        slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
    all_scores = tf.nn.softmax(chars_logit)
    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
    return ids, log_prob, scores 

def det_net_loss(seg_masks_in, reg_masks_in,
                 seg_preds, reg_preds,
                 reg_loss_weight=10.0,
                 epsilon=1e-5):

  with tf.variable_scope('loss'):
    out_size = seg_preds.get_shape()[1:3]
    seg_masks_in_ds = tf.image.resize_images(seg_masks_in[:,:,:,tf.newaxis],
                                             out_size[0], out_size[1],
                                             tf.image.ResizeMethod.NEAREST_NEIGHBOR)
    reg_masks_in_ds = tf.image.resize_images(reg_masks_in,
                                             out_size[0], out_size[1],
                                             tf.image.ResizeMethod.NEAREST_NEIGHBOR)

    # segmentation loss
    seg_masks_onehot = slim.one_hot_encoding(seg_masks_in_ds[:,:,:,0], 2)
    seg_loss = - tf.reduce_mean(seg_masks_onehot * tf.log(seg_preds + epsilon))

    # regression loss
    mask = tf.to_float(seg_masks_in_ds)
    reg_loss = tf.reduce_sum(mask * (reg_preds - reg_masks_in_ds)**2)
    reg_loss = reg_loss / (tf.reduce_sum(mask) + 1.0)

  return seg_loss + reg_loss_weight * reg_loss 

def test_create_summaries_is_runnable(self):
    ocr_model = self.create_model()
    data = data_provider.InputEndpoints(
        images=self.fake_images,
        images_orig=self.fake_images,
        labels=self.fake_labels,
        labels_one_hot=slim.one_hot_encoding(self.fake_labels,
                                             self.num_char_classes))
    endpoints = ocr_model.create_base(
        images=self.fake_images, labels_one_hot=None)
    charset = create_fake_charset(self.num_char_classes)
    summaries = ocr_model.create_summaries(
        data, endpoints, charset, is_training=False)
    with self.test_session() as sess:
      sess.run(tf.global_variables_initializer())
      sess.run(tf.local_variables_initializer())
      tf.tables_initializer().run()
      sess.run(summaries)  # just check it is runnable 

def char_predictions(self, chars_logit):
    """Returns confidence scores (softmax values) for predicted characters.

    Args:
      chars_logit: chars logits, a tensor with shape
        [batch_size x seq_length x num_char_classes]

    Returns:
      A tuple (ids, log_prob, scores), where:
        ids - predicted characters, a int32 tensor with shape
          [batch_size x seq_length];
        log_prob - a log probability of all characters, a float tensor with
          shape [batch_size, seq_length, num_char_classes];
        scores - corresponding confidence scores for characters, a float
        tensor
          with shape [batch_size x seq_length].
    """
    log_prob = utils.logits_to_log_prob(chars_logit)
    ids = tf.to_int32(tf.argmax(log_prob, dimension=2), name='predicted_chars')
    mask = tf.cast(
        slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
    all_scores = tf.nn.softmax(chars_logit)
    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
    return ids, log_prob, scores 

def test_create_summaries_is_runnable(self):
    ocr_model = self.create_model()
    data = data_provider.InputEndpoints(
        images=self.fake_images,
        images_orig=self.fake_images,
        labels=self.fake_labels,
        labels_one_hot=slim.one_hot_encoding(self.fake_labels,
                                             self.num_char_classes))
    endpoints = ocr_model.create_base(
        images=self.fake_images, labels_one_hot=None)
    charset = create_fake_charset(self.num_char_classes)
    summaries = ocr_model.create_summaries(
        data, endpoints, charset, is_training=False)
    with self.test_session() as sess:
      sess.run(tf.global_variables_initializer())
      sess.run(tf.local_variables_initializer())
      tf.tables_initializer().run()
      sess.run(summaries)  # just check it is runnable 

def char_predictions(self, chars_logit):
    """Returns confidence scores (softmax values) for predicted characters.

    Args:
      chars_logit: chars logits, a tensor with shape
        [batch_size x seq_length x num_char_classes]

    Returns:
      A tuple (ids, log_prob, scores), where:
        ids - predicted characters, a int32 tensor with shape
          [batch_size x seq_length];
        log_prob - a log probability of all characters, a float tensor with
          shape [batch_size, seq_length, num_char_classes];
        scores - corresponding confidence scores for characters, a float
        tensor
          with shape [batch_size x seq_length].
    """
    log_prob = utils.logits_to_log_prob(chars_logit)
    ids = tf.to_int32(tf.argmax(log_prob, dimension=2), name='predicted_chars')
    mask = tf.cast(
        slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
    all_scores = tf.nn.softmax(chars_logit)
    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
    return ids, log_prob, scores 

def test_create_summaries_is_runnable(self):
    ocr_model = self.create_model()
    data = data_provider.InputEndpoints(
        images=self.fake_images,
        images_orig=self.fake_images,
        labels=self.fake_labels,
        labels_one_hot=slim.one_hot_encoding(self.fake_labels,
                                             self.num_char_classes))
    endpoints = ocr_model.create_base(
        images=self.fake_images, labels_one_hot=None)
    charset = create_fake_charset(self.num_char_classes)
    summaries = ocr_model.create_summaries(
        data, endpoints, charset, is_training=False)
    with self.test_session() as sess:
      sess.run(tf.global_variables_initializer())
      sess.run(tf.local_variables_initializer())
      tf.tables_initializer().run()
      sess.run(summaries)  # just check it is runnable 

def fake_labels(batch_size, seq_length, num_char_classes):
  labels_np = tf.convert_to_tensor(
      np.random.randint(
          low=0, high=num_char_classes, size=(batch_size, seq_length)))
  return slim.one_hot_encoding(labels_np, num_classes=num_char_classes) 

def encode_coordinates_alt(self, net):
    """An alternative implemenation for the encoding coordinates.

    Args:
      net: a tensor of shape=[batch_size, height, width, num_features]

    Returns:
      a list of tensors with encoded image coordinates in them.
    """
    batch_size, h, w, _ = net.shape.as_list()
    h_loc = [
      tf.tile(
          tf.reshape(
              tf.contrib.layers.one_hot_encoding(
                  tf.constant([i]), num_classes=h), [h, 1]), [1, w])
      for i in xrange(h)
    ]
    h_loc = tf.concat([tf.expand_dims(t, 2) for t in h_loc], 2)
    w_loc = [
      tf.tile(
          tf.contrib.layers.one_hot_encoding(tf.constant([i]), num_classes=w),
          [h, 1]) for i in xrange(w)
    ]
    w_loc = tf.concat([tf.expand_dims(t, 2) for t in w_loc], 2)
    loc = tf.concat([h_loc, w_loc], 2)
    loc = tf.tile(tf.expand_dims(loc, 0), [batch_size, 1, 1, 1])
    return tf.concat([net, loc], 3) 

def char_one_hot(self, logit):
    """Creates one hot encoding for a logit of a character.

    Args:
      logit: A tensor with shape [batch_size, num_char_classes].

    Returns:
      A tensor with shape [batch_size, num_char_classes]
    """
    prediction = tf.argmax(logit, axis=1)
    return slim.one_hot_encoding(prediction, self._params.num_char_classes) 

def fake_labels(batch_size, seq_length, num_char_classes):
  labels_np = tf.convert_to_tensor(
      np.random.randint(
          low=0, high=num_char_classes, size=(batch_size, seq_length)))
  return slim.one_hot_encoding(labels_np, num_classes=num_char_classes) 

def char_one_hot(self, logit):
    """Creates one hot encoding for a logit of a character.

    Args:
      logit: A tensor with shape [batch_size, num_char_classes].

    Returns:
      A tensor with shape [batch_size, num_char_classes]
    """
    prediction = tf.argmax(logit, dimension=1)
    return slim.one_hot_encoding(prediction, self._params.num_char_classes) 

def encode_coordinates_alt(self, net):
    """An alternative implemenation for the encoding coordinates.

    Args:
      net: a tensor of shape=[batch_size, height, width, num_features]

    Returns:
      a list of tensors with encoded image coordinates in them.
    """
    batch_size, h, w, _ = net.shape.as_list()
    h_loc = [
      tf.tile(
          tf.reshape(
              tf.contrib.layers.one_hot_encoding(
                  tf.constant([i]), num_classes=h), [h, 1]), [1, w])
      for i in xrange(h)
    ]
    h_loc = tf.concat([tf.expand_dims(t, 2) for t in h_loc], 2)
    w_loc = [
      tf.tile(
          tf.contrib.layers.one_hot_encoding(tf.constant([i]), num_classes=w),
          [h, 1]) for i in xrange(w)
    ]
    w_loc = tf.concat([tf.expand_dims(t, 2) for t in w_loc], 2)
    loc = tf.concat([h_loc, w_loc], 2)
    loc = tf.tile(tf.expand_dims(loc, 0), [batch_size, 1, 1, 1])
    return tf.concat([net, loc], 3) 

def fake_labels(batch_size, seq_length, num_char_classes):
  labels_np = tf.convert_to_tensor(
      np.random.randint(
          low=0, high=num_char_classes, size=(batch_size, seq_length)))
  return slim.one_hot_encoding(labels_np, num_classes=num_char_classes) 

def encode_coordinates_alt(self, net):
    """An alternative implemenation for the encoding coordinates.

    Args:
      net: a tensor of shape=[batch_size, height, width, num_features]

    Returns:
      a list of tensors with encoded image coordinates in them.
    """
    batch_size, h, w, _ = net.shape.as_list()
    h_loc = [
      tf.tile(
          tf.reshape(
              tf.contrib.layers.one_hot_encoding(
                  tf.constant([i]), num_classes=h), [h, 1]), [1, w])
      for i in range(h)
    ]
    h_loc = tf.concat([tf.expand_dims(t, 2) for t in h_loc], 2)
    w_loc = [
      tf.tile(
          tf.contrib.layers.one_hot_encoding(tf.constant([i]), num_classes=w),
          [h, 1]) for i in range(w)
    ]
    w_loc = tf.concat([tf.expand_dims(t, 2) for t in w_loc], 2)
    loc = tf.concat([h_loc, w_loc], 2)
    loc = tf.tile(tf.expand_dims(loc, 0), [batch_size, 1, 1, 1])
    return tf.concat([net, loc], 3) 

def char_one_hot(self, logit):
    """Creates one hot encoding for a logit of a character.

    Args:
      logit: A tensor with shape [batch_size, num_char_classes].

    Returns:
      A tensor with shape [batch_size, num_char_classes]
    """
    prediction = tf.argmax(logit, dimension=1)
    return slim.one_hot_encoding(prediction, self._params.num_char_classes)