Torch7 Study Notes (iii) sequencialization

1. Serialization

Torch provides 4 advanced methods to serialize or deserialize arbitrary lua/torch objects. These methods are abstracted from the file object and created for ease of operation.

The first two methods are used to serialize or deserialize from a file:

Torch.save (Filename,object [, format])

[Object] torch.load (filename [, format])

The following two functions are serialized or deserialized from a string:

[STR] Torch.serialize (object [, format])

[Object] Torch.deserialize (str [, format])

Serialization into a file can be used to save any type of data structure, such as sharing. The advantage of serializing a string is that the data structure can be stored in a database, or in third-party software.

The usage of each function is explained in detail below.

1.1 Torch.save (Filename,object [, format])

Writes object objects to the file filename, format can be selected as ASCII and binary, and binary by default. Binary types are platform-dependent, but are compact and fast to read and write. The ASCII type is platform-independent and can be shared across platforms.

--Arbitrary object:

obj = {

Mat = Torch.randn (10,10),

Name = ' 10 ',

Test = {

Entry = 1

}

}

--Save to disk:

Torch.save (' Test.dat ', obj)

1.2 [Object] torch.load (filename [, format])

Reads the object from the file.

--Given serialized object from section above, Reload:

obj = torch.load (' Test.dat ')

Print (obj)

--Would print:

--{[Mat] = doubletensor-size:10x10

--[Name] = string: "10"

--[Test] = table-size:0}

1.3 [STR] Torch.serialize (object [, format])

Serializes an object into a string (

--Arbitrary object:

obj = {

Mat = Torch.randn (10,10),

Name = ' 10 ',

Test = {

Entry = 1

}

}

--Serialize:

str = torch.serialize (obj)

1.4 [Object] Torch.deserialize (str [, format])

Deserializes an object from a string, which is still binary by default.

--Given serialized object from section above, deserialize:

obj = torch.deserialize (str)

Print (obj)

--Would print:

--{[Mat] = doubletensor-size:10x10

--[Name] = string: "10"

--[Test] = table-size:0}

Torch7 Learning Note (iii) Sequencialization