Python keras.backend.variable() Examples

def build(self, input_shape):
		self.input_spec = [InputSpec(shape=input_shape)]
		self.input_dim = input_shape[2]

		self.W = self.init((self.output_dim, 4 * self.input_dim),
		                   name='{}_W'.format(self.name))
		self.U = self.inner_init((self.input_dim, 4 * self.input_dim),
		                         name='{}_U'.format(self.name))
		self.b = K.variable(np.hstack((np.zeros(self.input_dim),
		                               K.get_value(self.forget_bias_init((self.input_dim,))),
		                               np.zeros(self.input_dim),
		                               np.zeros(self.input_dim))),
		                    name='{}_b'.format(self.name))

		self.A = self.init((self.input_dim, self.output_dim),
		                    name='{}_A'.format(self.name))
		self.ba = K.zeros((self.output_dim,), name='{}_ba'.format(self.name))


		self.trainable_weights = [self.W, self.U, self.b, self.A, self.ba]

		if self.initial_weights is not None:
			self.set_weights(self.initial_weights)
			del self.initial_weights 

def __init__(self, lr=0.001, beta_1=0.9, beta_2=0.999,
                 epsilon=None, decay=0., amsgrad=False, accum_iters=1, **kwargs):
        if accum_iters < 1:
            raise ValueError('accum_iters must be >= 1')
        super(AdamAccumulate, self).__init__(**kwargs)
        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.lr = K.variable(lr, name='lr')
            self.beta_1 = K.variable(beta_1, name='beta_1')
            self.beta_2 = K.variable(beta_2, name='beta_2')
            self.decay = K.variable(decay, name='decay')
        if epsilon is None:
            epsilon = K.epsilon()
        self.epsilon = epsilon
        self.initial_decay = decay
        self.amsgrad = amsgrad
        self.accum_iters = K.variable(accum_iters, K.dtype(self.iterations))
        self.accum_iters_float = K.cast(self.accum_iters, K.floatx()) 

def get_total_loss(content_losses, style_losses, total_var_loss,
                   content_weights, style_weights, tv_weights, class_targets):
    total_loss = K.variable(0.)

    # Compute content losses
    for loss in content_losses:
        weighted_loss = K.mean(K.gather(content_weights, class_targets) * loss)
        weighted_content_losses.append(weighted_loss)
        total_loss += weighted_loss

    # Compute style losses
    for loss in style_losses:
        weighted_loss = K.mean(K.gather(style_weights, class_targets) * loss)
        weighted_style_losses.append(weighted_loss)
        total_loss += weighted_loss

    # Compute tv loss
    weighted_tv_loss = K.mean(K.gather(tv_weights, class_targets) *
                              total_var_loss)
    total_loss += weighted_tv_loss

    return (total_loss, weighted_content_losses, weighted_style_losses,
            weighted_tv_loss) 

def build(self, input_shape):
        # Create mean and count
        # These are weights because just maintaining variables don't get saved with the model, and we'd like
        # to have these numbers saved when we save the model.
        # But we need to make sure that the weights are untrainable.
        self.mean = self.add_weight(name='mean', 
                                      shape=input_shape[1:],
                                      initializer='zeros',
                                      trainable=False)
        self.count = self.add_weight(name='count', 
                                      shape=[1],
                                      initializer='zeros',
                                      trainable=False)

        # self.mean = K.zeros(input_shape[1:], name='mean')
        # self.count = K.variable(0.0, name='count')
        super(MeanStream, self).build(input_shape)  # Be sure to call this somewhere! 

def __init__(self, lr=0.001, final_lr=0.1, beta_1=0.9, beta_2=0.999, gamma=1e-3,
                 epsilon=None, decay=0., amsbound=False, weight_decay=0.0, **kwargs):
        super(AdaBound, self).__init__(**kwargs)

        if not 0. <= gamma <= 1.:
            raise ValueError("Invalid `gamma` parameter. Must lie in [0, 1] range.")

        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.lr = K.variable(lr, name='lr')
            self.beta_1 = K.variable(beta_1, name='beta_1')
            self.beta_2 = K.variable(beta_2, name='beta_2')
            self.decay = K.variable(decay, name='decay')

        self.final_lr = final_lr
        self.gamma = gamma

        if epsilon is None:
            epsilon = K.epsilon()
        self.epsilon = epsilon
        self.initial_decay = decay
        self.amsbound = amsbound

        self.weight_decay = float(weight_decay)
        self.base_lr = float(lr) 

def find_analogy_patches(a, a_prime, b, patch_size=3, patch_stride=1):
    '''This is for precalculating the analogy_loss

    Since A, A', and B never change we only need to calculate the patch matches once.
    '''
    # extract patches from feature maps
    a_patches, a_patches_norm = patches.make_patches(K.variable(a), patch_size, patch_stride)
    a_prime_patches, a_prime_patches_norm = patches.make_patches(K.variable(a_prime), patch_size, patch_stride)
    b_patches, b_patches_norm = patches.make_patches(K.variable(b), patch_size, patch_stride)
    # find best patches and calculate loss
    p = patches.find_patch_matches(b_patches, b_patches_norm, a_patches / a_patches_norm)
    #best_patches = a_prime_patches[p]
    best_patches = K.reshape(a_prime_patches[p], K.shape(b_patches))
    f = K.function([], best_patches)
    best_patches = f([])
    return best_patches 

def __init__(self, lr=1e-1, beta_1=0.9, beta_2=0.999,
                 epsilon=1e-8, decay=0., amsgrad=False, partial=1. / 8., **kwargs):
        if partial < 0 or partial > 0.5:
            raise ValueError(
                "Padam: 'partial' must be a positive float with a maximum "
                "value of `0.5`, since higher values will cause divergence "
                "during training."
            )
        super(Padam, self).__init__(**kwargs)
        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.lr = K.variable(lr, name='lr')
            self.beta_1 = K.variable(beta_1, name='beta_1')
            self.beta_2 = K.variable(beta_2, name='beta_2')
            self.decay = K.variable(decay, name='decay')
        if epsilon is None:
            epsilon = K.epsilon()
        self.epsilon = epsilon
        self.partial = partial
        self.initial_decay = decay
        self.amsgrad = amsgrad 

def __init__(self,
                 lr,
                 momentum=0.9,
                 weight_decay=0.0001,
                 eeta=0.001,
                 epsilon=0.0,
                 nesterov=False,
                 **kwargs):

        if momentum < 0.0:
            raise ValueError("momentum should be positive: %s" % momentum)
        if weight_decay < 0.0:
            raise ValueError("weight_decay is not positive: %s" % weight_decay)
        super(LARS, self).__init__(**kwargs)
        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.lr = K.variable(lr, name='lr')
            self.momentum = K.variable(momentum, name='momentum')
            self.weight_decay = K.variable(weight_decay, name='weight_decay')
            self.eeta = K.variable(eeta, name='eeta')
        self.epsilon = epsilon
        self.nesterov = nesterov 

def __init__(self, lr=0.01, beta_1=0.9, beta_2=0.999,
                 epsilon=1e-3, decay=0., **kwargs):
        super(Yogi, self).__init__(**kwargs)
        if beta_1 <= 0 or beta_1 >= 1:
            raise ValueError("beta_1 has to be in ]0, 1[")
        if beta_2 <= 0 or beta_2 >= 1:
            raise ValueError("beta_2 has to be in ]0, 1[")

        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.lr = K.variable(lr, name='lr')
            self.beta_1 = K.variable(beta_1, name='beta_1')
            self.beta_2 = K.variable(beta_2, name='beta_2')
            self.decay = K.variable(decay, name='decay')
        if epsilon is None:
            epsilon = K.epsilon()
        if epsilon <= 0:
            raise ValueError("epsilon has to be larger than 0")
        self.epsilon = epsilon
        self.initial_decay = decay 

def test_sub_pixel_upscaling(scale_factor):
    num_samples = 2
    num_row = 16
    num_col = 16
    input_dtype = K.floatx()

    nb_channels = 4 * (scale_factor ** 2)
    input_data = np.random.random((num_samples, nb_channels, num_row, num_col))
    input_data = input_data.astype(input_dtype)

    if K.image_data_format() == 'channels_last':
        input_data = input_data.transpose((0, 2, 3, 1))

    input_tensor = K.variable(input_data)
    expected_output = K.eval(KC.depth_to_space(input_tensor,
                                               scale=scale_factor))

    layer_test(SubPixelUpscaling,
               kwargs={'scale_factor': scale_factor},
               input_data=input_data,
               expected_output=expected_output,
               expected_output_dtype=K.floatx()) 

def check_composed_tensor_operations(first_function_name, first_function_args,
                                     second_function_name, second_function_args,
                                     input_shape):
    ''' Creates a random tensor t0 with shape input_shape and compute
                 t1 = first_function_name(t0, **first_function_args)
                 t2 = second_function_name(t1, **second_function_args)
        with both Theano and TensorFlow backends and ensures the answers match.
    '''
    val = np.random.random(input_shape) - 0.5
    xth = KTH.variable(val)
    xtf = KTF.variable(val)

    yth = getattr(KCTH, first_function_name)(xth, **first_function_args)
    ytf = getattr(KCTF, first_function_name)(xtf, **first_function_args)

    zth = KTH.eval(getattr(KCTH, second_function_name)(yth, **second_function_args))
    ztf = KTF.eval(getattr(KCTF, second_function_name)(ytf, **second_function_args))

    assert zth.shape == ztf.shape
    assert_allclose(zth, ztf, atol=1e-05) 

def __init__(self, s=3, skip=True):
        self.skip = skip
        self.s = K.variable(s, name='s_constraint') 

def __init__(self, s=3, skip=True):
        self.skip = skip
        self.s = K.variable(s, name='s_constraint') 

def __init__(self):
        self.gamma = K.variable(2.) 

def yolo_eval(yolo_outputs,
              image_shape,
              max_boxes=10,
              score_threshold=.6,
              iou_threshold=.5):
    """Evaluate YOLO model on given input batch and return filtered boxes."""
    box_xy, box_wh, box_confidence, box_class_probs = yolo_outputs
    boxes = yolo_boxes_to_corners(box_xy, box_wh)
    boxes, scores, classes = yolo_filter_boxes(
        boxes, box_confidence, box_class_probs, threshold=score_threshold)

    # Scale boxes back to original image shape.
    height = image_shape[0]
    width = image_shape[1]
    image_dims = K.stack([height, width, height, width])
    image_dims = K.reshape(image_dims, [1, 4])
    boxes = boxes * image_dims

    # TODO: Something must be done about this ugly hack!
    max_boxes_tensor = K.variable(max_boxes, dtype='int32')
    K.get_session().run(tf.variables_initializer([max_boxes_tensor]))
    nms_index = tf.image.non_max_suppression(
        boxes, scores, max_boxes_tensor, iou_threshold=iou_threshold)
    boxes = K.gather(boxes, nms_index)
    scores = K.gather(scores, nms_index)
    classes = K.gather(classes, nms_index)
    return boxes, scores, classes 

def build(self, input_shape):
        self.input_spec = [InputSpec(shape=input_shape)]
        shape = (int(input_shape[self.axis]),)

        self.gamma = K.variable(self.gamma_init(shape), name='%s_gamma' % self.name)
        self.beta = K.variable(self.beta_init(shape), name='%s_beta' % self.name)
        self.trainable_weights = [self.gamma, self.beta]

        if self.initial_weights is not None:
            self.set_weights(self.initial_weights)
            del self.initial_weights 

def build(self, input_shape):
        self.input_spec = [InputSpec(shape=input_shape)]
        shape = (int(input_shape[self.axis]),)

        self.gamma = K.variable(self.gamma_init(shape), name='%s_gamma' % self.name)
        self.beta = K.variable(self.beta_init(shape), name='%s_beta' % self.name)
        self.trainable_weights = [self.gamma, self.beta]

        if self.initial_weights is not None:
            self.set_weights(self.initial_weights)
            del self.initial_weights 

def __init__(self, learning_rate=0.001, beta_1=0.9, beta_2=0.999,
                 amsgrad=False, model=None, zero_penalties=True,
                 batch_size=32, total_iterations=0, total_iterations_wd=None,
                 use_cosine_annealing=False, lr_multipliers=None,
                 weight_decays=None, init_verbose=True,
                 eta_min=0, eta_max=1, t_cur=0, **kwargs):
        if total_iterations > 1:
            weight_decays = _init_weight_decays(model, zero_penalties,
                                                weight_decays)

        self.initial_decay = kwargs.pop('decay', 0.0)
        self.epsilon = kwargs.pop('epsilon', K.epsilon())
        learning_rate = kwargs.pop('lr', learning_rate)
        eta_t = kwargs.pop('eta_t', 1.)
        super(AdamW, self).__init__(**kwargs)

        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.learning_rate = K.variable(learning_rate, name='learning_rate')
            self.beta_1 = K.variable(beta_1, name='beta_1')
            self.beta_2 = K.variable(beta_2, name='beta_2')
            self.decay = K.variable(self.initial_decay, name='decay')
            self.eta_min = K.constant(eta_min, name='eta_min')
            self.eta_max = K.constant(eta_max, name='eta_max')
            self.eta_t = K.variable(eta_t, dtype='float32', name='eta_t')
            self.t_cur = K.variable(t_cur, dtype='int64', name='t_cur')

        self.batch_size = batch_size
        self.total_iterations = total_iterations
        self.total_iterations_wd = total_iterations_wd or total_iterations
        self.amsgrad = amsgrad
        self.lr_multipliers = lr_multipliers
        self.weight_decays = weight_decays or {}
        self.init_verbose = init_verbose
        self.use_cosine_annealing = use_cosine_annealing

        _check_args(self, total_iterations, use_cosine_annealing, weight_decays)
        self._init_lr = learning_rate  # to print lr_mult setup
        self._init_notified = False 

def __init__(self, learning_rate=0.002, beta_1=0.9, beta_2=0.999,
                 model=None, zero_penalties=True, batch_size=32,
                 total_iterations=0, total_iterations_wd=None,
                 use_cosine_annealing=False, lr_multipliers=None,
                 weight_decays=None, init_verbose=True,
                 eta_min=0, eta_max=1, t_cur=0, **kwargs):
        if total_iterations > 1:
            weight_decays = _init_weight_decays(model, zero_penalties,
                                                weight_decays)

        self.schedule_decay = kwargs.pop('schedule_decay', 0.004)
        self.epsilon = kwargs.pop('epsilon', K.epsilon())
        learning_rate = kwargs.pop('lr', learning_rate)
        eta_t = kwargs.pop('eta_t', 1.)
        super(NadamW, self).__init__(**kwargs)

        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.m_schedule = K.variable(1., name='m_schedule')
            self.learning_rate = K.variable(learning_rate, name='learning_rate')
            self.beta_1 = K.variable(beta_1, name='beta_1')
            self.beta_2 = K.variable(beta_2, name='beta_2')
            self.eta_min = K.constant(eta_min, name='eta_min')
            self.eta_max = K.constant(eta_max, name='eta_max')
            self.eta_t = K.variable(eta_t, dtype='float32', name='eta_t')
            self.t_cur = K.variable(t_cur, dtype='int64', name='t_cur')

        self.batch_size = batch_size
        self.total_iterations = total_iterations
        self.total_iterations_wd = total_iterations_wd or total_iterations
        self.lr_multipliers = lr_multipliers
        self.weight_decays = weight_decays or {}
        self.use_cosine_annealing = use_cosine_annealing
        self.init_verbose = init_verbose

        _check_args(self, total_iterations, use_cosine_annealing, weight_decays)
        self._init_lr = learning_rate  # to print lr_mult setup
        self._init_notified = False 

def __init__(self, learning_rate=0.01, momentum=0., nesterov=False,
                 model=None, zero_penalties=True, batch_size=32,
                 total_iterations=0, total_iterations_wd=None,
                 use_cosine_annealing=False, lr_multipliers=None,
                 weight_decays=None, init_verbose=True,
                 eta_min=0, eta_max=1, t_cur=0, **kwargs):
        if total_iterations > 1:
            weight_decays = _init_weight_decays(model, zero_penalties,
                                                weight_decays)

        self.initial_decay = kwargs.pop('decay', 0.0)
        learning_rate = kwargs.pop('lr', learning_rate)
        eta_t = kwargs.pop('eta_t', 1.)
        super(SGDW, self).__init__(**kwargs)

        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.learning_rate = K.variable(learning_rate, name='learning_rate')
            self.momentum = K.variable(momentum, name='momentum')
            self.decay = K.variable(self.initial_decay, name='decay')
            self.eta_min = K.constant(eta_min, name='eta_min')
            self.eta_max = K.constant(eta_max, name='eta_max')
            self.eta_t = K.variable(eta_t, dtype='float32', name='eta_t')
            self.t_cur = K.variable(t_cur, dtype='int64', name='t_cur')

        self.batch_size = batch_size
        self.total_iterations = total_iterations
        self.total_iterations_wd = total_iterations_wd or total_iterations
        self.nesterov = nesterov
        self.lr_multipliers = lr_multipliers
        self.weight_decays = weight_decays or {}
        self.init_verbose = init_verbose
        self.use_cosine_annealing = use_cosine_annealing

        _check_args(self, total_iterations, use_cosine_annealing, weight_decays)
        self._init_lr = learning_rate  # to print lr_mult setup
        self._init_notified = False 

def __init__(self, lr=0.001, beta_1=0.9, beta_2=0.999, amsgrad=False,
                 epsilon=None, decay=0.0, model=None, zero_penalties=True,
                 batch_size=32, total_iterations=0, total_iterations_wd=None,
                 use_cosine_annealing=False, lr_multipliers=None,
                 weight_decays=None, init_verbose=True,
                 eta_min=0, eta_max=1, t_cur=0, **kwargs):
        if total_iterations > 1:
            weight_decays = _init_weight_decays(model, zero_penalties,
                                                weight_decays)
        eta_t = kwargs.pop('eta_t', 1.)
        super(AdamW, self).__init__(**kwargs)

        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.lr = K.variable(lr, name='lr')
            self.beta_1 = K.variable(beta_1, name='beta_1')
            self.beta_2 = K.variable(beta_2, name='beta_2')
            self.decay = K.variable(decay, name='decay')
            self.eta_min = K.constant(eta_min, name='eta_min')
            self.eta_max = K.constant(eta_max, name='eta_max')
            self.eta_t = K.variable(eta_t, dtype='float32', name='eta_t')
            self.t_cur = K.variable(t_cur, dtype='int64', name='t_cur')

        self.initial_decay = decay
        self.epsilon = epsilon or K.epsilon()
        self.batch_size = batch_size
        self.total_iterations = total_iterations
        self.total_iterations_wd = total_iterations_wd or total_iterations
        self.amsgrad = amsgrad
        self.lr_multipliers = lr_multipliers
        self.weight_decays = weight_decays or {}
        self.init_verbose = init_verbose
        self.use_cosine_annealing = use_cosine_annealing

        _check_args(self, total_iterations, use_cosine_annealing, weight_decays)
        self._init_lr = lr  # to print lr_mult setup
        self._init_notified = False 

def __init__(self, lr=0.002, beta_1=0.9, beta_2=0.999,
                 schedule_decay=0.004, epsilon=None,
                 model=None, zero_penalties=True, batch_size=32,
                 total_iterations=0, total_iterations_wd=None,
                 use_cosine_annealing=False, lr_multipliers=None,
                 weight_decays=None, init_verbose=True,
                 eta_min=0, eta_max=1, t_cur=0, **kwargs):
        if total_iterations > 1:
            weight_decays = _init_weight_decays(model, zero_penalties,
                                                weight_decays)
        eta_t = kwargs.pop('eta_t', 1.)
        super(NadamW, self).__init__(**kwargs)

        with K.name_scope(self.__class__.__name__):
            self.iterations = K.variable(0, dtype='int64', name='iterations')
            self.m_schedule = K.variable(1., name='m_schedule')
            self.lr = K.variable(lr, name='lr')
            self.beta_1 = K.variable(beta_1, name='beta_1')
            self.beta_2 = K.variable(beta_2, name='beta_2')
            self.eta_min = K.constant(eta_min, name='eta_min')
            self.eta_max = K.constant(eta_max, name='eta_max')
            self.eta_t = K.variable(eta_t, dtype='float32', name='eta_t')
            self.t_cur = K.variable(t_cur, dtype='int64', name='t_cur')

        self.epsilon = epsilon or K.epsilon()
        self.schedule_decay = schedule_decay
        self.batch_size = batch_size
        self.total_iterations = total_iterations
        self.total_iterations_wd = total_iterations_wd or total_iterations
        self.lr_multipliers = lr_multipliers
        self.weight_decays = weight_decays or {}
        self.use_cosine_annealing = use_cosine_annealing
        self.init_verbose = init_verbose

        _check_args(self, total_iterations, use_cosine_annealing, weight_decays)
        self._init_lr = lr  # to print lr_mult setup
        self._init_notified = False 

def build(self, input_shape):
        self.input_spec = [InputSpec(shape=input_shape)]
        shape = (int(input_shape[self.axis]),)

        # Compatibility with TensorFlow >= 1.0.0
        self.gamma = K.variable(self.gamma_init(shape), name='{}_gamma'.format(self.name))
        self.beta = K.variable(self.beta_init(shape), name='{}_beta'.format(self.name))
        #self.gamma = self.gamma_init(shape, name='{}_gamma'.format(self.name))
        #self.beta = self.beta_init(shape, name='{}_beta'.format(self.name))
        self.trainable_weights = [self.gamma, self.beta]

        if self.initial_weights is not None:
            self.set_weights(self.initial_weights)
            del self.initial_weights 

def yolo_non_max_suppression(scores, boxes, classes, max_boxes = 10, iou_threshold = 0.5):
    max_boxes_tensor = K.variable(max_boxes, dtype='int32') # tensor to be used in tf.image.non_max_suppression()
    K.get_session().run(tf.variables_initializer([max_boxes_tensor])) # initialize variable max_boxes_tensor
    
    # Use tf.image.non_max_suppression() to get the list of indices corresponding to boxes you keep
    nms_indices = tf.image.non_max_suppression(boxes, scores, max_boxes, iou_threshold)
    
    # Use K.gather() to select only nms_indices from scores, boxes and classes
    scores = K.gather(scores, nms_indices)
    boxes = K.gather(boxes, nms_indices)
    classes = K.gather(classes, nms_indices)
    
    return scores, boxes, classes 

def yolo_non_max_suppression(scores, boxes, classes, max_boxes = 10, iou_threshold = 0.5):
    max_boxes_tensor = K.variable(max_boxes, dtype='int32') # tensor to be used in tf.image.non_max_suppression()
    K.get_session().run(tf.variables_initializer([max_boxes_tensor])) # initialize variable max_boxes_tensor
    
    # Use tf.image.non_max_suppression() to get the list of indices corresponding to boxes you keep
    nms_indices = tf.image.non_max_suppression(boxes, scores, max_boxes, iou_threshold)
    
    # Use K.gather() to select only nms_indices from scores, boxes and classes
    scores = K.gather(scores, nms_indices)
    boxes = K.gather(boxes, nms_indices)
    classes = K.gather(classes, nms_indices)
    
    return scores, boxes, classes 

def yolo_eval(yolo_outputs,
              image_shape,
              max_boxes=10,
              score_threshold=.6,
              iou_threshold=.5):
    """Evaluate YOLO model on given input batch and return filtered boxes."""
    box_confidence, box_xy, box_wh, box_class_probs = yolo_outputs
    boxes = yolo_boxes_to_corners(box_xy, box_wh)
    boxes, scores, classes = yolo_filter_boxes(
        box_confidence, boxes, box_class_probs, threshold=score_threshold)
    
    # Scale boxes back to original image shape.
    height = image_shape[0]
    width = image_shape[1]
    image_dims = K.stack([height, width, height, width])
    image_dims = K.reshape(image_dims, [1, 4])
    boxes = boxes * image_dims

    # TODO: Something must be done about this ugly hack!
    max_boxes_tensor = K.variable(max_boxes, dtype='int32')
    K.get_session().run(tf.variables_initializer([max_boxes_tensor]))
    nms_index = tf.image.non_max_suppression(
        boxes, scores, max_boxes_tensor, iou_threshold=iou_threshold)
    boxes = K.gather(boxes, nms_index)
    scores = K.gather(scores, nms_index)
    classes = K.gather(classes, nms_index)
    
    return boxes, scores, classes 

def mean_normal(shape, mean=1., scale=0.02, name=None):
    return K.variable(np.random.normal(loc=mean, scale=scale, size=shape), name=name) 

def mean_normal(shape, mean=1., scale=0.02, name=None):
    return K.variable(np.random.normal(loc=mean, scale=scale, size=shape), name=name) 

def build(self, input_shape):
		self.input_spec = [InputSpec(shape=input_shape)]
		input_dim = input_shape[2]
		self.input_dim = input_dim
		
		if self.stateful:
			self.reset_states()
		else:
			self.states = [None, None]
			self.states_dim = [self.input_dim, self.output_dim]


		self.weight_size = self.output_dim * 4
		self.W = self.add_weight((input_dim, self.weight_size),
                                 initializer=self.init,
                                 name='{}_W'.format(self.name),
                                 regularizer=self.W_regularizer)
		self.U = self.add_weight((input_dim, self.weight_size),
                                 initializer=self.inner_init,
                                 name='{}_U'.format(self.name),
                                 regularizer=self.U_regularizer)

		def b_reg(shape, name=None):
			return K.variable(np.hstack((np.zeros(self.output_dim),
										K.get_value(self.forget_bias_init((self.output_dim,))),
										np.zeros(self.output_dim),
										np.zeros(self.output_dim))),
										name='{}_b'.format(self.name))
		self.b = self.add_weight((self.weight_size,),
                                     initializer=b_reg,
                                     name='{}_b'.format(self.name),
                                     regularizer=self.b_regularizer)


		if self.initial_weights is not None:
			self.set_weights(self.initial_weights)
			del self.initial_weights

		self.built = True 

def test_keras_unstack_hack():
    y_true_np = np.random.random([1, 3, 2])
    y_true_np[:, :, 0] = 0
    y_true_np[:, :, 1] = 1

    y_true_keras = K.variable(y_true_np)

    y, u = wtte._keras_unstack_hack(y_true_keras)
    y_true_keras_new = K.stack([y, u], axis=-1)

    np.testing.assert_array_equal(K.eval(y_true_keras_new), y_true_np)

# SANITY CHECK: Use pure Weibull data censored at C(ensoring point).
# Should converge to the generating A(alpha) and B(eta) for each timestep