If you have the full version of TensorFlow installed, you can leverage tensorflow.contrib.learn.

from sklearn import datasets, metrics
from tensorflow.contrib import skflow

run_config = skflow.estimators.RunConfig(num_cores=3)

iris = datasets.load_iris()
classifier = learn.TensorFlowLinearClassifier(n_classes=3,steps=800, config=run_config)
classifier.fit(iris.data, iris.target)
score = metrics.accuracy_score(iris.target, classifier.predict(iris.data))
print("Accuracy: %f" % score)

# This will give you the following output... note, it's
# a complete model.

# Step #100, epoch #20, avg. train loss: 0.56387
# Step #200, epoch #40, avg. train loss: 0.39505
# Step #300, epoch #60, avg. train loss: 0.33969
# Step #400, epoch #80, avg. train loss: 0.30646
# Step #500, epoch #100, avg. train loss: 0.28121
# Step #600, epoch #120, avg. train loss: 0.26077
# Step #700, epoch #140, avg. train loss: 0.24579
# Step #800, epoch #160, avg. train loss: 0.23332
# Accuracy: 0.973333
# In [27]: