There's this utility which has on print_tensors_in_checkpoint_file method http://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/tools/inspect_checkpoint.py>

或者，您可以使用 Saver 来恢复模型并在变量张量上使用 session.run 以获取 numpy 数组形式的值

I'd like to read the weights and visualize them as images. But I don't see any documentation about model format and how to read the trained weights.

There's this utility which has on print_tensors_in_checkpoint_file method http://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/tools/inspect_checkpoint.py

Alternatively, you can use Saver to restore the model and use session.run on variable tensors to get values as numpy arrays