marker.py 文件源码

def sample2vec(self, sample_arr):
        # ???????????0?ids  1?label ids
        ut_arr = []
        batch_size = len(sample_arr)
        labels = np.zeros((batch_size, self.label_size))
        vec_cache = []
        for i in range(batch_size):
            pad = [data_utils.PAD_ID]*(self.max_ut_size-len(sample_arr[i][0]))  #0???
            vec_cache.append(list(reversed(sample_arr[i][0]+pad)))  #????
            for j in range(len(sample_arr[i][1])):
                index = int (sample_arr[i][1][j])
                if index < self.label_size:
                    labels[i][index] = 1.0
        for i in range(self.max_ut_size):
            temp = np.array([ vec_cache[index][i] for index in range(batch_size)])
            ut_arr.append(temp)
        #?????id???lable???
        return ut_arr, labels

#######################################################