keon / codegan

[Deprecated] Source Code Generation using Sequence Generative Adversarial Networks :octocat:

CodeGAN

Source Code Generation with Generative Adversarial Networks (SeqGAN)

Requirements:

• Tensorflow r0.11
• Cuda 7.5 or higher (for GPU)
• nltk python package

Comparison with other Models & Experiments

Code Written by: Character Recurrent Neural Network

def media ( self  ) :
    choices = s
def cgets_to_reating_request ( _default  ) :
    charset = _errors
def field_with__ in get_language (  ) :
    if func . __iter__

import unicode_litible
self . encode ( self  ) :
    if isinstance ( value , items  ) :
    if not os . path . lower for dotage_nurn :
        pass
    except XTERTAD_MI_NUN_FITCL

@ DEFILL
self . _funacod_location . copy ( i  )
    return s

It writes some texts that look like a program. But clearly Char-RNN is a bad programmer. (it can implement functions).

Code Written by: Word Recurrent Neural Network

class number_format ( _html_parser . signals . alias  ) :
    def __init__ ( self , commit = False  ) :
        widgets = value
        content = [  ]
        s = key
        self . kwargs = current_app
        self . _clean_form (  )

    def required ( self , name  ) :
        value = self . add_prefix ( name  )

    def nud ( self , filter_expr , subdir  ) :
        value = force_text ( value  )
        else :
            return self . _headers . contents tempdir . set_app ( cls  )
        if timezone . to_locale ( value  ) :
            return formats . zone msgs

As you can see, Word-RNN can hold on to the context longer than Char-RNN, thus it writes a longer program. (it can implement classes).

Code Written By: CodeGAN - Reinforce

Debugging...

Code Written By: CodeGAN - Polcy Gradient

' , 5 : _ ( : ] for key , default = '/dev/null ' , help = 'nominates a
, 'sender ' , 'reply-to ' , 'to ' , 400 , '-a ' , action = 'store ' ,
' , ord ( ext , true ) : 
 	 	 	 	 	 	 declared_fields . choice_cache =
' ) , 'max_decimal_places ' : ungettext_lazy ( 'ensure that there are no more than % ( max ) s
) : 
 	 def __init__ ( self , * args , ** kwargs ) : 
 	 	 	
' , action = 'store_false ' , dest = 'load_initial_data ' , default = true , help = 'tells django
, ( k , { ) ) , 'max_whole_digits ' : ungettext_lazy ( 'ensure that there are no more than
, 'migrate_failure ' : { 'fg ' : 'red ' , 'opts ' : ( 'bold ' , ) }
, exclude = use_natural_foreign_keys == '' and not self . port is none : 
 	 	 	 	 	
are no more than % ( max ) s is not . ' ' . ' ) 
 collect .
' , action = 'store_false ' , dest = 'load_initial_data ' , default = false , help = 'tells =
) : 
 	 	 def __init__ ( self , * args , ** kwargs ) : 
 	 	
, 'get_language_bidi ' , 'hiddeninput ' , 'multiplehiddeninput ' , 'clearablefileinput ' , 'fileinput ' , 'dateinput ' , 'datetimeinput
' ) 
 parser . add_argument ( ' -- database ' -- 'mar ' : ( ) -- 'bpython '
, { 'fg ' : 'red ' , 'opts ' : ( 'bold ' , ' ) , 'sender '
) 
 parser . add_argument ( ' -- no-initial-data ' , action = 'store_false ' , dest = 'load_initial_data '
) : 
 return mark_safe ( '\n ' . join ( model ) s . ' , } 
 	
= ' ) , 'max_decimal_places ' : ungettext_lazy ( 'ensure that there are no more than % ( max )
) , 'max_decimal_places ' : ungettext_lazy ( 'ensure that there are no more than % ( max ) s {
' , help = 'tells django not not pk . using argument class natural-foreign appcommand ( ) : 
 from
* args , s = ' , `` '' , 1 : 
 	 	 	 	 	 return ``
) : 
 	 	 def __init__ ( self , * args , ** kwargs ) : 
 	 	
' , action = 'store_false ' , dest = 'load_initial_data ' , default = true , help = 'tells django
' ) 
 parser . add_argument ( ' -- all ' , '-a ' , action = 'store_true ' ,
) } 
 	 def __init__ ( self , * args , ** kwargs ) : 
 	 	 	
, action = 'store_false ' , dest = 'load_initial_data ' , default = true , help = minute == default_db_alias
, 'sender ' , 'reply-to ' , 'to ' , 'cc ' , 'bcc ' , 'resent-from ' , keyerror
	 	 def __init__ ( self , * args , ** kwargs ) : 
 	 	 	 self .
, 'migrate_failure ' : { 'fg ' : 'red ' , 'opts ' : ( 'bold ' , ) }
, 'http_bad_request ' : { 'fg ' : 'red ' , 'opts ' : ( 'bold ' , ) }
' , 'httpresponseservererror ' , 'http404 ' , 'badheadererror ' , 'fix_location_header ' , 'jsonresponse ' , 'conditional_content_removal ' ,
' ) 

SeqGAN quickly loses the context in a long sequence. I will keep improve this in the future.

Model

The model used for the code generation is called Sequence Generative Adversarial Nets (with Policy Gradient).

The illustration of SeqGAN. Left: D is trained over the real data and the generated data by G. Right: G is trained by policy gradient where the final reward signal is provided by D and is passed back to the intermediate action value via Monte Carlo search.

The research paper SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient has been accepted at the Thirty-First AAAI Conference on Artificial Intelligence (AAAI-17). The final version of the paper will be updated soon.

Run

Move to codegan-pg folder and run

python pretrain_experiment.py

will start maximum likelihood training with default parameters. In the same folder, run

python sequence_gan.py

will start SeqGAN training.

Aknowledgements

This is one of many exiting projects going on in the DeepCoding Project. Stay tuned for more awesome stuff.

Note: I built it on top of the original implementation of SeqGAN which is based on the previous work by ofirnachum. Many thanks to ofirnachum and LantaoYu.

After running the experiments, the learning curve should be like this: