Fluctuation in multiple runs

[HanyinWang]

Thanks for the great repo.

I was trying the run the Graph CNN model multiple times and got fluctuations between the runs.

I tried to fix the following seed:

random_seed = 99
tf.random.set_random_seed(random_seed)
random.seed(random_seed)
np.random.seed(random_seed)
os.environ['PYTHONHASHSEED']=str(random_seed)

As well as the following steps:
indices = np.random.RandomState(seed=random_seed).permutation(A.nnz // 2)[:n]
rows = np.random.RandomState(seed=random_seed).randint(0, M, n)
cols = np.random.RandomState(seed=random_seed).randint(0, M, n)
vals = np.random.RandomState(seed=random_seed).uniform(0, 1, n)

I'm still experiencing fluctuations between runs.

Could you please kindly provide the solution on how to fix the performance between runs?

Thanks,
Hanyin