def __init__(
self, shape, filter_size, num_features, forget_bias=1.0, input_size=None, state_is_tuple=False,
act=tf.nn.tanh
):
"""Initialize the basic Conv LSTM cell."""
# if not state_is_tuple:
# logging.warn("%s: Using a concatenated state is slower and will soon be "
# "deprecated. Use state_is_tuple=True.", self)
if input_size is not None:
logging.warn("%s: The input_size parameter is deprecated.", self)
self.shape = shape
self.filter_size = filter_size
self.num_features = num_features
self._forget_bias = forget_bias
self._state_is_tuple = state_is_tuple
self._activation = act
def __init__(
self, shape, filter_size, num_features, forget_bias=1.0, input_size=None, state_is_tuple=False,
act=tf.nn.tanh
):
"""Initialize the basic Conv LSTM cell."""
# if not state_is_tuple:
# logging.warn("%s: Using a concatenated state is slower and will soon be "
# "deprecated. Use state_is_tuple=True.", self)
if input_size is not None:
logging.warn("%s: The input_size parameter is deprecated.", self)
self.shape = shape
self.filter_size = filter_size
self.num_features = num_features
self._forget_bias = forget_bias
self._state_is_tuple = state_is_tuple
self._activation = act
def __init__(self, network, idx, quant_func, dataset):
logging.info("### init DeQuant_Layer")
self.type = EL_DEQUANTIZE
self.typename = "EL_DEQUANTIZE"
layer = network.all_layers[idx]
shape = layer._nodes[0].out_tensors[0].shape
self.count = get_tensor_size(shape)
if (self.count > 256 * 1024):
logging.warn(
"output>1MB data, we assume it is dbg usage, cut to 1MB")
self.count = 256 * 1024
minv, maxv, _ = quant_func(network, layer, dataset)
self.scale, self.bias = min_max_to_scale_bias(minv, maxv)
self.memsize = self.count * (4 + 1)
self.outsize = self.count * 4
logging.info("###dequant layer: count=%d, sclale=%f, bias=%f" %
(self.count, self.scale, self.bias))