Reproducibility of a ML project?

I am working on tensorflow, numpy for some NLP projects. I can't reproduce results though I have used tensorflow and numpy random seed.

I have 4 *.py file.

Then I do the following things blindly but it's also not working.

At the beginning of the each of the file I have added the following code,

import tensorflow as tf
tf.set_random_seed(1234)
import numpy as np
np.random.seed(1000)

Is there anyting I need to do more?

Note: I have also used Adversarial Learning for the project. [I'm not sure if this is relevant or not to this problem.]