To print out a text representation of the current graph:

tf.train.export_meta_graph(@"sharp.meta.txt", as_text:true);

Doing this in both TensorFlow.NET and Python allows you to compare the graph nodes with a text diffing tool.

To visualize the TensorFlow.NET-graph with Tensorboard, first export it as binary meta file:

tf.train.export_meta_graph(@"sharp.meta", as_text:false);

Then use Python to convert the meta format into an event file:

import tensorflow
tf=tensorflow

saver = tf.train.import_meta_graph("sharp.meta")
writer = tf.summary.FileWriter(logdir="c:/tensorboard/logdir", graph=tf.get_default_graph())  # write to event
writer.flush()

Start Tensorboard:

tensorboard --logdir C:\tensorboard\logdir

Which will visualize the graph nicely.

The above graph is produced by this little pice of code:

var graph = tf.Graph().as_default();
with<Graph>(graph, g =>
{
	var x = constant_op.constant(10);
	var true_fn = new Func<Tensor>(() =>
	{
	    var (c_op, op_desc) = ops._create_c_op(g, ops._NodeDef("Identity", "cond/myop"), new[] { x }, new Operation[0]);
	    return x;
	});
	control_flow_ops.cond(x < 10, true_fn, () => x);
});

Doing the same in Python is much easier, we can directly write the event file:

writer = tf.summary.FileWriter(logdir="D:/dev/tensorboard/logdir", graph=tf.get_default_graph())  # write to event
writer.flush()

Comparing the viszalized graphs can make finding bugs a lot easier.