Question

我正在阅读用于文本分类的CNN模型code link，我在第70行想知道代码：

b = tf.Variable(tf.constant(0.1, shape=[num_classes]), name="b")

为什么它可以同时定义为变量和常量？这等于：

b = tf.Variable(0.1, shape=[num_classes], name="b")

Answer 1

是的，两者都是一样的。 Tensorflow隐式将tf.constant值复制到tf.Variable值。操作a.op，b.op和c.op解释一切

  import tensorflow as tf

    with tf.Session() as sess:
        a=tf.constant(0.1);
        b = tf.Variable(tf.constant(0.1), name="b");
        c = tf.Variable(0.1, name="b");
        sess.run(tf.global_variables_initializer());
        print(a.dtype);
        print(b.dtype);
        print(c.dtype);
        print("**********************")
        print(a.op);
        print(b.op);
        print(c.op);

输出：

<dtype: 'float32'>
<dtype: 'float32_ref'>
<dtype: 'float32_ref'>
**********************

name: "Const_40"
op: "Const"
attr {
  key: "dtype"
  value {
    type: DT_FLOAT
  }
}
attr {
  key: "value"
  value {
    tensor {
      dtype: DT_FLOAT
      tensor_shape {
      }
      float_val: 0.10000000149
    }
  }
}

name: "b_38"
op: "VariableV2"
attr {
  key: "container"
  value {
    s: ""
  }
}
attr {
  key: "dtype"
  value {
    type: DT_FLOAT
  }
}
attr {
  key: "shape"
  value {
    shape {
    }
  }
}
attr {
  key: "shared_name"
  value {
    s: ""
  }
}

name: "b_39"
op: "VariableV2"
attr {
  key: "container"
  value {
    s: ""
  }
}
attr {
  key: "dtype"
  value {
    type: DT_FLOAT
  }
}
attr {
  key: "shape"
  value {
    shape {
    }
  }
}
attr {
  key: "shared_name"
  value {
    s: ""
  }
}

tf.Variable和tf.constant

1 个答案: