pickle_data.py 文件源码

def pickle_transitions_matrix_data():
    transitions = pickle.load( open( "/ssd/ddimitrov/pickle/transitions", "rb" ) )
    vocab = pickle.load( open( "/ssd/ddimitrov/pickle/vocab", "rb" ) )

    i_indices = array.array(str("l"))
    j_indices = array.array(str("l"))
    values = array.array(str("d"))

    for s, targets in transitions.iteritems():
        for t, v in targets.iteritems():
            i_indices.append(vocab[s])
            j_indices.append(vocab[t])
            values.append(v)

    i_indices = np.frombuffer(i_indices, dtype=np.int_)
    j_indices = np.frombuffer(j_indices, dtype=np.int_)
    values = np.frombuffer(values, dtype=np.float64)
    transition_matrix=[i_indices,j_indices,values]
    pickle.dump(transition_matrix, open("/ssd/ddimitrov/pickle/transition_matrix", "wb"), protocol=pickle.HIGHEST_PROTOCOL)
    print "transition_matrix"