Tensorflow tflearn initializations uniform
tflearn.initializations.uniform
Uniform
tflearn.initializations.uniform (shape=None, minval=0, maxval=None, dtype=tf.float32, seed=None)
Initialization with random values from a uniform distribution.
The generated values follow a uniform distribution in the range [minval, maxval). The lower bound minval is included in the range, while the upper bound maxval is excluded.
For floats, the default range is [0, 1). For ints, at least maxval must be specified explicitly.
In the integer case, the random integers are slightly biased unless maxval - minval is an exact power of two. The bias is small for values of maxval - minval significantly smaller than the range of the output (either 232 or 264).
- Arguments
shape: List of int. A shape to initialize a Tensor (optional). dtype: The tensor data type. Only float are supported. seed: int. Used to create a random seed for the distribution.
- Returns The Initializer, or an initialized Tensor if shape is specified.