sess run tensorflow

1tf.compat.v1.Session
2A class for running TensorFlow operations.
3
4run(
5    fetches, feed_dict=None, options=None, run_metadata=None
6)
7
8Runs operations and evaluates tensors in fetches.
9
10This method runs one "step" of TensorFlow computation, 
11by running the necessary graph fragment to execute every Operation 
12and evaluate every Tensor in fetches, substituting the values in 
13feed_dict for the corresponding input values.
14
15The fetches argument may be a single graph element, or an arbitrarily 
16nested list, tuple, namedtuple, dict, or OrderedDict containing graph 
17elements at its leaves.
18
19The optional feed_dict argument allows the caller to override the 
20value of tensors in the graph. Each value in feed_dict must be 
21convertible to a numpy array of the dtype of the corresponding key.
22
23The optional options argument expects a [RunOptions] proto. 
24The options allow controlling the behavior of this particular step 
25(e.g. turning tracing on).
26
27The optional run_metadata argument expects a [RunMetadata] proto. 
28When appropriate, the non-Tensor output of this step will be 
29collected there. For example, when users turn on tracing in options, 
30the profiled info will be collected into this argument and passed 
31back.