参考语义角色标注的代码

def load_parameter(file_name, h, w):
    with open(file_name, 'rb') as f:
        f.read(16)  # skip header.
        return np.fromfile(f, dtype=np.float32).reshape(h, w)

解析通过fluid.io.save_inference_model(save_dirname, test_feed_order, [scale_infer], exe)保存的词向量term_table

    query_basic_emb = layers.embedding(
        input=query_basic,
        dtype='float32',
        size=[1000002, 128],
        param_attr='term_table',
        is_sparse=IS_SPARSE)

报错，看起来是header的大小不对，其中a是保存的二进制term_table路径，看起来f.read(28)就可以成功

>>> load_parameter(a, 1000002, 128)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "load_embedding.py", line 25, in load_parameter
    return np.fromfile(f, dtype=np.float32).reshape(h, w)
ValueError: cannot reshape array of size 128000259 into shape (1000002,128)