WOLFRAM

Wolfram Language & System Documentation Center
Wolfram Language Home Page »

NetStateObject
✖
NetStateObject

[Experimental]

✖

creates an object derived from net that represents a neural net with additional stored state information that is updated when the net is applied to data.

✖
NetStateObject[net,seed]

creates an object in which additional stored state information is initialized using seed.

Details

  • NetStateObject[…][data] updates stored state information in the NetStateObject.
  • State information is associated with the state ports of recurrent net layers such as LongShortTermMemoryLayer.
  • NetStateObject will not store the state of layers whose state ports are initialized from other layers in a NetGraph.
  • When a seed is not provided, initial values for recurrent states will consist of arrays of zeros.
  • The current value of the stored states is given by NetExtract[NetStateObject[…],"States"].

Examples

open allclose all

Basic Examples  (3)Summary of the most common use cases

Create a recurrent net:

In[1]:=1
✖
https://wolfram.com/xid/0btogq8kt8a6-lenhyi
Out[1]=1

Create a state object from the recurrent net:

In[2]:=2
✖
https://wolfram.com/xid/0btogq8kt8a6-jo9f6l
Out[2]=2

Evaluate the state object on some data:

In[3]:=3
✖
https://wolfram.com/xid/0btogq8kt8a6-q3dvup
Out[3]=3

Due to the presence of the stored states, the behavior of the state object can change between evaluations, even on the same input:

In[4]:=4
✖
https://wolfram.com/xid/0btogq8kt8a6-6tvvja
Out[4]=4
In[5]:=5
✖
https://wolfram.com/xid/0btogq8kt8a6-w5m1cq
Out[5]=5

Create a state object with a specified initial state:

In[1]:=1
✖
https://wolfram.com/xid/0btogq8kt8a6-6r4d84
Out[1]=1

Create a classifier that predicts the next element of a sequence:

In[1]:=1
✖
https://wolfram.com/xid/0btogq8kt8a6-50km1e
Out[1]=1

Train the classifier on a set of input sequences:

In[2]:=2
✖
https://wolfram.com/xid/0btogq8kt8a6-11hss3
Out[2]=2

Create a state object and use it to efficiently generate a maximum-likelihood sequence, starting from a single 1:

In[3]:=3
✖
https://wolfram.com/xid/0btogq8kt8a6-frwekw
Out[3]=3
In[4]:=4
✖
https://wolfram.com/xid/0btogq8kt8a6-1vpktq
Out[4]=4

Applications  (1)Sample problems that can be solved with this function

Training an English character-level language model. First, create 300,000 training examples of 25 characters each from two novels:

In[1]:=1
✖
https://wolfram.com/xid/0btogq8kt8a6-mvbupv
In[2]:=2
✖
https://wolfram.com/xid/0btogq8kt8a6-1kl129

The data is of the form of a classification problem: given a sequence of characters, predict the next one. A sample of the data:

In[3]:=3
✖
https://wolfram.com/xid/0btogq8kt8a6-8jfew7

Obtain the list of all characters in the text:

In[4]:=4
✖
https://wolfram.com/xid/0btogq8kt8a6-7280lz
Out[4]=4

Define a net that takes in a string of characters and returns a prediction for the next character:

In[5]:=5
✖
https://wolfram.com/xid/0btogq8kt8a6-xklmy7

Train the net. This can take up to an hour on a CPU; it is recommended set option TargetDevice to "GPU" if a supported GPU is available:

In[6]:=6
✖
https://wolfram.com/xid/0btogq8kt8a6-h5r4f5
Out[6]=6

Predict the next character, given a sequence of characters:

In[7]:=7
✖
https://wolfram.com/xid/0btogq8kt8a6-qpr8fs
Out[7]=7

Generate 100 characters of text, given starting text:

In[8]:=8
✖
https://wolfram.com/xid/0btogq8kt8a6-0u6153
In[9]:=9
✖
https://wolfram.com/xid/0btogq8kt8a6-8365bs
Out[9]=9

One can get more interesting text by sampling from the probability distribution of predictions:

In[10]:=10
✖
https://wolfram.com/xid/0btogq8kt8a6-4ln4uh
In[11]:=11
✖
https://wolfram.com/xid/0btogq8kt8a6-1xrxrq
Out[11]=11

Properties & Relations  (3)Properties of the function, and connections to other functions

If the initial value of a recurrent layer's state is provided by a connection in a NetGraph, that state will not be stored by NetStateObject.

Create a graph that uses a connection to provide the initial value of the state of a BasicRecurrentLayer:

In[1]:=1
✖
https://wolfram.com/xid/0btogq8kt8a6-kvurdg
Out[1]=1

This graph cannot be used inside a NetStateObject, as there are no states left to store:

In[2]:=2
✖
https://wolfram.com/xid/0btogq8kt8a6-5ajeac
Out[2]=2

The current value of the stored states can be obtained using NetExtract.

First create a NetStateObject:

In[1]:=1
✖
https://wolfram.com/xid/0btogq8kt8a6-71as95
Out[1]=1

Apply the object to some data:

In[2]:=2
✖
https://wolfram.com/xid/0btogq8kt8a6-7bp3fz
Out[2]=2

Extract the current value of the states:

In[3]:=3
✖
https://wolfram.com/xid/0btogq8kt8a6-scfztm
Out[3]=3

For recurrent nets, using a NetStateObject is equivalent to manually keeping track of the recurrent states via NetPort[All,"States"].

To see this, create a classifier that predicts the next element of a sequence:

In[1]:=1
✖
https://wolfram.com/xid/0btogq8kt8a6-nn7f9i
Out[1]=1

Train the classifier on a set of input sequences:

In[2]:=2
✖
https://wolfram.com/xid/0btogq8kt8a6-ddl6e0
Out[2]=2

Create a state object and use it to efficiently generate a maximum-likelihood sequence, starting from a single 1:

In[3]:=3
✖
https://wolfram.com/xid/0btogq8kt8a6-h2y8gj
Out[3]=3
In[4]:=4
✖
https://wolfram.com/xid/0btogq8kt8a6-65uvwt
Out[4]=4

Generate from the trained net using NetPort[All,"States"] to set and get the recurrent states, which yields the same result:

In[5]:=5
✖
https://wolfram.com/xid/0btogq8kt8a6-itejp8
Out[5]=5
Wolfram Research (2018), NetStateObject, Wolfram Language function, https://reference.wolfram.com/language/ref/NetStateObject.html.
✖
Wolfram Research (2018), NetStateObject, Wolfram Language function, https://reference.wolfram.com/language/ref/NetStateObject.html.

Text

Wolfram Research (2018), NetStateObject, Wolfram Language function, https://reference.wolfram.com/language/ref/NetStateObject.html.

✖
Wolfram Research (2018), NetStateObject, Wolfram Language function, https://reference.wolfram.com/language/ref/NetStateObject.html.

CMS

Wolfram Language. 2018. "NetStateObject." Wolfram Language & System Documentation Center. Wolfram Research. https://reference.wolfram.com/language/ref/NetStateObject.html.

✖
Wolfram Language. 2018. "NetStateObject." Wolfram Language & System Documentation Center. Wolfram Research. https://reference.wolfram.com/language/ref/NetStateObject.html.

APA

Wolfram Language. (2018). NetStateObject. Wolfram Language & System Documentation Center. Retrieved from https://reference.wolfram.com/language/ref/NetStateObject.html

✖
Wolfram Language. (2018). NetStateObject. Wolfram Language & System Documentation Center. Retrieved from https://reference.wolfram.com/language/ref/NetStateObject.html

BibTeX

@misc{reference.wolfram_2025_netstateobject, author="Wolfram Research", title="{NetStateObject}", year="2018", howpublished="\url{https://reference.wolfram.com/language/ref/NetStateObject.html}", note=[Accessed: 26-March-2025 ]}

✖
@misc{reference.wolfram_2025_netstateobject, author="Wolfram Research", title="{NetStateObject}", year="2018", howpublished="\url{https://reference.wolfram.com/language/ref/NetStateObject.html}", note=[Accessed: 26-March-2025 ]}

BibLaTeX

@online{reference.wolfram_2025_netstateobject, organization={Wolfram Research}, title={NetStateObject}, year={2018}, url={https://reference.wolfram.com/language/ref/NetStateObject.html}, note=[Accessed: 26-March-2025 ]}

✖
@online{reference.wolfram_2025_netstateobject, organization={Wolfram Research}, title={NetStateObject}, year={2018}, url={https://reference.wolfram.com/language/ref/NetStateObject.html}, note=[Accessed: 26-March-2025 ]}