Class: Tensorflow::Session

Inherits:
Object
  • Object
show all
Defined in:
lib/tensorflow/session.rb

Overview

A class for running TensorFlow operations. A Session object encapsulates the environment in which Operation objects are executed, and Tensor objects are evaluated. A Session instance lets a caller drive a TensorFlow graph computation. When a Session is created with a given target, a new Session object is bound to the universe of resources specified by that target. Those resources are available to this session to perform computation described in the GraphDef. After extending the session with a graph, the caller uses the Run() API to perform the computation and potentially fetch outputs as Tensors. Protocol buffer exposes various configuration options for a session. The Op definations are stored in ops.pb file. Official documentation of session and Operation.

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(graph, c_options = nil) ⇒ Session

A TensorFlow graph is a description of computations. To compute anything, a graph must be launched in a Session. A Session places the graph ops and provides methods to execute them.



19
20
21
22
23
24
25
26
27
 
# File 'lib/tensorflow/session.rb', line 19

def initialize(graph, c_options = nil)
    c_options = Tensorflow::Session_options.new if c_options.nil?
    self.status = Tensorflow::Status.new
    cOpt = c_options.c
    c_session = Tensorflow::TF_NewSession(graph.c, cOpt, status.c)
    Tensorflow::TF_DeleteSessionOptions(cOpt)
    raise 'Error in session initialization.' if status.code != 0
    self.c = c_session
end

Instance Attribute Details

#c ⇒ Object

Returns the value of attribute c.



11
12
13
 
# File 'lib/tensorflow/session.rb', line 11

def c
  @c
end

#dimensions ⇒ Object

Create a success status.



19
20
21
22
23
24
25
26
27
 
# File 'lib/tensorflow/session.rb', line 19

def initialize(graph, c_options = nil)
    c_options = Tensorflow::Session_options.new if c_options.nil?
    self.status = Tensorflow::Status.new
    cOpt = c_options.c
    c_session = Tensorflow::TF_NewSession(graph.c, cOpt, status.c)
    Tensorflow::TF_DeleteSessionOptions(cOpt)
    raise 'Error in session initialization.' if status.code != 0
    self.c = c_session
end

#graph ⇒ Object

A TensorFlow graph is a description of computations. To compute anything, a graph must be launched in a Session. A Session places the graph ops and provides methods to execute them.



19
20
21
 
# File 'lib/tensorflow/session.rb', line 19

def graph
  @graph
end

#ops ⇒ Object

Nodes in the graph are called ops (short for operations). An op takes zero or more Tensors, performs some computation, and produces zero or more Tensors.



19
20
21
 
# File 'lib/tensorflow/session.rb', line 19

def ops
  @ops
end

#session ⇒ Object

Returns the value of attribute session.



11
12
13
 
# File 'lib/tensorflow/session.rb', line 11

def session
  @session
end

#status ⇒ Object

Returns the value of attribute status.



11
12
13
 
# File 'lib/tensorflow/session.rb', line 11

def status
  @status
end

Instance Method Details

#close ⇒ Object 



70
71
72
73
74
75
76
77
 
# File 'lib/tensorflow/session.rb', line 70

def close
    status = Tensorflow::Status.new
    Tensorflow::TF_CloseSession(c, status.c)
    raise 'Error in closing session.' if status.code != 0
    Tensorflow::TF_DeleteSession(c, status.c)
    raise 'Error in deleting session.' if status.code != 0
    self.c = nil
end

#run(inputs, outputs, targets) ⇒ Object

Run the graph with the associated session starting with the supplied feeds to compute the value of the requested fetches. Runs, but does not return Tensors for operations specified in targets.

On success, returns the fetched Tensors in the same order as supplied in the fetches argument. If fetches is set to nil, the returned Tensor fetches is empty.

Note that the caller maintains responsibility for the input tensors, so the caller should still call tensor.delete() on each input tensor



39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
 
# File 'lib/tensorflow/session.rb', line 39

def run(inputs, outputs, targets)
    inputPorts = Tensorflow::TF_Output_vector.new
    inputValues = Tensorflow::Tensor_Vector.new
    inputs.each do |port, tensor|
        inputPorts.push(port.c)
        inputValues.push(tensor.tensor)
    end

    outputPorts = Tensorflow::TF_Output_vector.new
    outputs.each do |output|
        outputPorts.push(output.c)
    end

    cTargets = Tensorflow::TF_Operation_vector.new
    targets.each do |targe|
        cTargets.push(targe.c)
    end

    outputValues = Tensorflow::Session_run(c, inputPorts, inputValues, outputPorts, cTargets)

    output_array = []
    outputValues.each do |value|
        converted_value = convert_value_for_output_array(value)
        output_array.push(converted_value)
        # Since we're returning the results as an array, there's no point keeping
        # the output tensor, so we delete it. 
        Tensorflow::TF_DeleteTensor(value)
    end
    output_array
end