tf.random.Generator

Random-number generator.

tf.random.Generator(
    copy_from=None, state=None, alg=None
)

Example:

Creating a generator from a seed:

g = tf.random.Generator.from_seed(1234)
g.normal(shape=(2, 3))
<tf.Tensor: shape=(2, 3), dtype=float32, numpy=
array([[ 0.9356609 ,  1.0854305 , -0.93788373],
       [-0.5061547 ,  1.3169702 ,  0.7137579 ]], dtype=float32)>

Creating a generator from a non-deterministic state:

g = tf.random.Generator.from_non_deterministic_state()
g.normal(shape=(2, 3))
<tf.Tensor: shape=(2, 3), dtype=float32, numpy=...>

All the constructors allow explicitly choosing an Random-Number-Generation (RNG) algorithm. Supported algorithms are "philox" and "threefry". For example:

g = tf.random.Generator.from_seed(123, alg="philox")
g.normal(shape=(2, 3))
<tf.Tensor: shape=(2, 3), dtype=float32, numpy=
array([[ 0.8673864 , -0.29899067, -0.9310337 ],
       [-1.5828488 ,  1.2481191 , -0.6770643 ]], dtype=float32)>

CPU, GPU and TPU with the same algorithm and seed will generate the same integer random numbers. Float-point results (such as the output of normal) may have small numerical discrepancies between different devices.

This class uses a tf.Variable to manage its internal state. Every time random numbers are generated, the state of the generator will change. For example:

g = tf.random.Generator.from_seed(1234)
g.state
<tf.Variable ... numpy=array([1234,    0,    0])>
g.normal(shape=(2, 3))
<...>
g.state
<tf.Variable ... numpy=array([2770,    0,    0])>

The shape of the state is algorithm-specific.

There is also a global generator:

g = tf.random.get_global_generator()
g.normal(shape=(2, 3))
<tf.Tensor: shape=(2, 3), dtype=float32, numpy=...>

When creating a generator inside a tf.distribute.Strategy scope, each replica will get a different stream of random numbers.

For example, in this code:

strat = tf.distribute.MirroredStrategy(devices=["cpu:0", "cpu:1"])
with strat.scope():
  g = tf.random.Generator.from_seed(1)
  def f():
    return g.normal([])
  results = strat.run(f).values

results[0] and results[1] will have different values.

If the generator is seeded (e.g. created via Generator.from_seed), the random numbers will be determined by the seed, even though different replicas get different numbers. One can think of a random number generated on a replica as a hash of the replica ID and a "master" random number that may be common to all replicas. Hence, the whole system is still deterministic.

(Note that the random numbers on different replicas are not correlated, even if they are deterministically determined by the same seed. They are not correlated in the sense that no matter what statistics one calculates on them, there won't be any discernable correlation.)

Generators can be freely saved and restored using tf.train.Checkpoint. The checkpoint can be restored in a distribution strategy with a different number of replicas than the original strategy. If a replica ID is present in both the original and the new distribution strategy, its state will be properly restored (i.e. the random-number stream from the restored point will be the same as that from the saving point) unless the replicas have already diverged in their RNG call traces before saving (e.g. one replica has made one RNG call while another has made two RNG calls). We don't have such guarantee if the generator is saved in a strategy scope and restored outside of any strategy scope, or vice versa.

copy_from a generator to be copied from.
state a vector of dtype STATE_TYPE representing the initial state of the RNG, whose length and semantics are algorithm-specific. If it's a variable, the generator will reuse it instead of creating a new variable.
alg the RNG algorithm. Possible values are tf.random.Algorithm.PHILOX for the Philox algorithm and tf.random.Algorithm.THREEFRY for the ThreeFry algorithm (see paper 'Parallel Random Numbers: As Easy as 1, 2, 3' [https://www.thesalmons.org/john/random123/papers/random123sc11.pdf]). The string names "philox" and "threefry" can also be used. Note PHILOX guarantees the same numbers are produced (given the same random state) across all architectures (CPU, GPU, XLA etc).

algorithm The RNG algorithm id (a Python integer or scalar integer Tensor).
key The 'key' part of the state of a counter-based RNG.

For a counter-base RNG algorithm such as Philox and ThreeFry (as described in paper 'Parallel Random Numbers: As Easy as 1, 2, 3' [https://www.thesalmons.org/john/random123/papers/random123sc11.pdf]), the RNG state consists of two parts: counter and key. The output is generated via the formula: output=hash(key, counter), i.e. a hashing of the counter parametrized by the key. Two RNGs with two different keys can be thought as generating two independent random-number streams (a stream is formed by increasing the counter).

state The internal state of the RNG.

Methods

binomial

View source

binomial(
    shape, counts, probs, dtype=tf.dtypes.int32, name=None
)

Outputs random values from a binomial distribution.

The generated values follow a binomial distribution with specified count and probability of success parameters.

Example:

counts = [10., 20.]
# Probability of success.
probs = [0.8]

rng = tf.random.Generator.from_seed(seed=234)
binomial_samples = rng.binomial(shape=[2], counts=counts, probs=probs)


counts = ... # Shape [3, 1, 2]
probs = ...  # Shape [1, 4, 2]
shape = [3, 4, 3, 4, 2]
rng = tf.random.Generator.from_seed(seed=1717)
# Sample shape will be [3, 4, 3, 4, 2]
binomial_samples = rng.binomial(shape=shape, counts=counts, probs=probs)

Args
shape A 1-D integer Tensor or Python array. The shape of the output tensor.
counts Tensor. The counts of the binomial distribution. Must be broadcastable with probs, and broadcastable with the rightmost dimensions of shape.
probs Tensor. The probability of success for the binomial distribution. Must be broadcastable with counts and broadcastable with the rightmost dimensions of shape.
dtype The type of the output. Default: tf.int32
name A name for the operation (optional).

Returns
samples A Tensor of the specified shape filled with random binomial values. For each i, each samples[i, ...] is an independent draw from the binomial distribution on counts[i] trials with probability of success probs[i].

from_key_counter

View source

@classmethod
from_key_counter(
    key, counter, alg
)

Creates a generator from a key and a counter.

This constructor only applies if the algorithm is a counter-based algorithm. See method key for the meaning of "key" and "counter".

Args
key the key for the RNG, a scalar of type STATE_TYPE.
counter a vector of dtype STATE_TYPE representing the initial counter for the RNG, whose length is algorithm-specific.,
alg the RNG algorithm. If None, it will be auto-selected. See __init__ for its possible values.

Returns
The new generator.

from_non_deterministic_state

View source

@classmethod
from_non_deterministic_state(
    alg=None
)

Creates a generator by non-deterministically initializing its state.

The source of the non-determinism will be platform- and time-dependent.

Args
alg (optional) the RNG algorithm. If None, it will be auto-selected. See __init__ for its possible values.

Returns
The new generator.

from_seed

View source

@classmethod
from_seed(
    seed, alg=None
)

Creates a generator from a seed.

A seed is a 1024-bit unsigned integer represented either as a Python integer or a vector of integers. Seeds shorter than 1024-bit will be padded. The padding, the internal structure of a seed and the way a seed is converted to a state are all opaque (unspecified). The only semantics specification of seeds is that two different seeds are likely to produce two independent generators (but no guarantee).

Args
seed the seed for the RNG.
alg (optional) the RNG algorithm. If None, it will be auto-selected. See __init__ for its possible values.

Returns
The new generator.

from_state

View source

@classmethod
from_state(
    state, alg
)

Creates a generator from a state.

See __init__ for description of state and alg.

Args
state the new state.
alg the RNG algorithm.

Returns
The new generator.

make_seeds

View source

make_seeds(
    count=1
)

Generates seeds for stateless random ops.

For example:

seeds = get_global_generator().make_seeds(count=10)
for i in range(10):
  seed = seeds[:, i]
  numbers = stateless_random_normal(shape=[2, 3], seed=seed)
  ...

Args
count the number of seed pairs (note that stateless random ops need a pair of seeds to invoke).

Returns
A tensor of shape [2, count] and dtype int64.

normal

View source

normal(
    shape, mean=0.0, stddev=1.0, dtype=tf.dtypes.float32, name=None
)

Outputs random values from a normal distribution.

Args
shape A 1-D integer Tensor or Python array. The shape of the output tensor.
mean A 0-D Tensor or Python value of type dtype. The mean of the normal distribution.
stddev A 0-D Tensor or Python value of type dtype. The standard deviation of the normal distribution.
dtype The type of the output.
name A name for the operation (optional).

Returns
A tensor of the specified shape filled with random normal values.

reset

View source

reset(
    state
)

Resets the generator by a new state.

See __init__ for the meaning of "state".

Args
state the new state.

reset_from_key_counter

View source

reset_from_key_counter(
    key, counter
)

Resets the generator by a new key-counter pair.

See from_key_counter for the meaning of "key" and "counter".

Args
key the new key.
counter the new counter.

reset_from_seed

View source

reset_from_seed(
    seed
)

Resets the generator by a new seed.

See from_seed for the meaning of "seed".

Args
seed the new seed.

skip

View source

skip(
    delta
)

Advance the counter of a counter-based RNG.

Args
delta the amount of advancement. The state of the RNG after skip(n) will be the same as that after normal([n]) (or any other distribution). The actual increment added to the counter is an unspecified implementation detail.

Returns
A Tensor of type int64.

split

View source

split(
    count=1
)

Returns a list of independent Generator objects.

Two generators are independent of each other in the sense that the random-number streams they generate don't have statistically detectable correlations. The new generators are also independent of the old one. The old generator's state will be changed (like other random-number generating methods), so two calls of split will return different new generators.

For example:

gens = get_global_generator().split(count=10)
for gen in gens:
  numbers = gen.normal(shape=[2, 3])
  # ...
gens2 = get_global_generator().split(count=10)
# gens2 will be different from gens

The new generators will be put on the current device (possible different from the old generator's), for example:

with tf.device("/device:CPU:0"):
  gen = Generator(seed=1234)  # gen is on CPU
with tf.device("/device:GPU:0"):
  gens = gen.split(count=10)  # gens are on GPU

Args
count the number of generators to return.

Returns
A list (length count) of Generator objects independent of each other. The new generators have the same RNG algorithm as the old one.

truncated_normal

View source

truncated_normal(
    shape, mean=0.0, stddev=1.0, dtype=tf.dtypes.float32, name=None
)

Outputs random values from a truncated normal distribution.

The generated values follow a normal distribution with specified mean and standard deviation, except that values whose magnitude is more than 2 standard deviations from the mean are dropped and re-picked.

Args
shape A 1-D integer Tensor or Python array. The shape of the output tensor.
mean A 0-D Tensor or Python value of type dtype. The mean of the truncated normal distribution.
stddev A 0-D Tensor or Python value of type dtype. The standard deviation of the normal distribution, before truncation.
dtype The type of the output.
name A name for the operation (optional).

Returns
A tensor of the specified shape filled with random truncated normal values.

uniform

View source

uniform(
    shape, minval=0, maxval=None, dtype=tf.dtypes.float32, name=None
)

Outputs 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 float numbers especially low-precision types like bfloat16, because of rounding, the result may sometimes include maxval.)

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 2**32 or 2**64).

For full-range random integers, pass minval=None and maxval=None with an integer dtype (for integer dtypes, minval and maxval must be both None or both not None).

Args
shape A 1-D integer Tensor or Python array. The shape of the output tensor.
minval A Tensor or Python value of type dtype, broadcastable with shape (for integer types, broadcasting is not supported, so it needs to be a scalar). The lower bound (included) on the range of random values to generate. Pass None for full-range integers. Defaults to 0.
maxval A Tensor or Python value of type dtype, broadcastable with shape (for integer types, broadcasting is not supported, so it needs to be a scalar). The upper bound (excluded) on the range of random values to generate. Pass None for full-range integers. Defaults to 1 if dtype is floating point.
dtype The type of the output.
name A name for the operation (optional).

Returns
A tensor of the specified shape filled with random uniform values.

Raises
ValueError If dtype is integral and maxval is not specified.

uniform_full_int

View source

uniform_full_int(
    shape, dtype=tf.dtypes.uint64, name=None
)

Uniform distribution on an integer type's entire range.

This method is the same as setting minval and maxval to None in the uniform method.

Args
shape the shape of the output.
dtype (optional) the integer type, default to uint64.
name (optional) the name of the node.

Returns
A tensor of random numbers of the required shape.