Beispiel #1
0
    def __init__(self,
                 corpus,
                 env_matrix,
                 context_type='sentence',
                 psi=None,
                 rand_perm=None,
                 lmda=7):
        """    
        Initialize BeagleOrderMulti.

        :param corpus: Source of observed data.
        :type corpus: Corpus

        :param env_matrix: BEAGLE environement matrix.
        :type env_matrix: 2-D array

        :param context_type: Name of tokenization stored in `corpus` whose
            tokens will be treated as documents. Default is `sentence`.
        :type context_type: string, optional

        :param psi:  
        :type psi: int, optional

        :param rand_perm:  
        :type rand_perm: boolean, optional

        :param lmda:  
        :type lmda: int, optional
        """
        self.context_type = context_type
        self.sents = corpus.view_contexts(context_type)
        self.dtype = env_matrix.dtype

        global _shape
        _shape = mp.Array('i', 2, lock=False)
        _shape[:] = env_matrix.shape

        print('Copying env matrix to shared mp array')
        global _env_matrix
        _env_matrix = mp.Array('d', env_matrix.size, lock=False)
        _env_matrix[:] = env_matrix.ravel()[:]

        #TODO: Convert this to a shared data structure
        global _b_conv
        _b_conv = mk_b_conv(env_matrix.shape[1], rand_perm)

        global _psi
        _psi = mp.Array('d', _shape[1], lock=False)
        if psi is None:
            _psi[:] = rand_pt_unit_sphere(env_matrix.shape[1])[:]
        else:
            _psi[:] = psi[:]

        global _lmda
        _lmda = mp.Value('i', lock=False)
        _lmda = lmda
Beispiel #2
0
    def __init__(self, corpus, env_matrix, context_type='sentence',
                 psi=None, rand_perm=None, lmda =7):
        """    
        Initialize BeagleOrderMulti.

        :param corpus: Source of observed data.
        :type corpus: Corpus

        :param env_matrix: BEAGLE environement matrix.
        :type env_matrix: 2-D array

        :param context_type: Name of tokenization stored in `corpus` whose
            tokens will be treated as documents. Default is `sentence`.
        :type context_type: string, optional

        :param psi:  
        :type psi: int, optional

        :param rand_perm:  
        :type rand_perm: boolean, optional

        :param lmda:  
        :type lmda: int, optional
        """
        self.context_type = context_type
        self.sents = corpus.view_contexts(context_type)
        self.dtype = env_matrix.dtype

        global _shape 
        _shape = mp.Array('i', 2, lock=False)
        _shape[:] = env_matrix.shape

        print('Copying env matrix to shared mp array')
        global _env_matrix
        _env_matrix = mp.Array('d', env_matrix.size, lock=False)
        _env_matrix[:] = env_matrix.ravel()[:]

        #TODO: Convert this to a shared data structure
        global _b_conv 
        _b_conv = mk_b_conv(env_matrix.shape[1], rand_perm)

        global _psi        
        _psi = mp.Array('d', _shape[1], lock=False)
        if psi is None:
            _psi[:] = rand_pt_unit_sphere(env_matrix.shape[1])[:]
        else:
            _psi[:] = psi[:]

        global _lmda
        _lmda = mp.Value('i', lock=False)
        _lmda =lmda
Beispiel #3
0
    def __init__(self,
                 corpus,
                 env_matrix,
                 context_type='sentence',
                 psi=None,
                 rand_perm=None,
                 lmda=7):
        """ 
        Initialize BeagleOrderSeq.
        
        :param corpus: Soure of observed data.
        :type corpus: Corpus

        :param env_matrix: BEAGLE environment matrix.
        :type env_matrix: 2-D array

        :param context_type: Name of tokenization stored in `corpus` whose
            tokens will be treated as documents. Default is `sentence`.
        :type context_type: string, optional

        :param psi:  
        :type psi: int, optional

        :param rand_perm:  
        :type rand_perm: boolean, optional

        :param lmda:  
        :type lmda: int, optional
        """
        self.context_type = context_type
        self.sents = corpus.view_contexts(context_type)
        self.env_matrix = env_matrix

        self.b_conv = mk_b_conv(env_matrix.shape[1], rand_perm)

        if psi is None:
            self.psi = rand_pt_unit_sphere(env_matrix.shape[1])
        else:
            self.psi = psi

        self.lmda = lmda
Beispiel #4
0
    def __init__(self, corpus, env_matrix, context_type='sentence',
                 psi=None, rand_perm=None, lmda =7):
        """ 
        Initialize BeagleOrderSeq.
        
        :param corpus: Soure of observed data.
        :type corpus: Corpus

        :param env_matrix: BEAGLE environment matrix.
        :type env_matrix: 2-D array

        :param context_type: Name of tokenization stored in `corpus` whose
            tokens will be treated as documents. Default is `sentence`.
        :type context_type: string, optional

        :param psi:  
        :type psi: int, optional

        :param rand_perm:  
        :type rand_perm: boolean, optional

        :param lmda:  
        :type lmda: int, optional
        """
        self.context_type = context_type
        self.sents = corpus.view_contexts(context_type)
        self.env_matrix = env_matrix

        self.b_conv = mk_b_conv(env_matrix.shape[1], rand_perm)
        
        if psi is None:
            self.psi = rand_pt_unit_sphere(env_matrix.shape[1])
        else:
            self.psi = psi

        self.lmda = lmda