from keras.models import Model
from keras import layers, regularizers
from keras.layers import Input
import logging
import ltns.logictensornetworks as ltn
logger = logging.getLogger()
logger.basicConfig = logging.basicConfig(level=logging.DEBUG)
import ltns.logictensornetworks_wrapper as ltnw
import tensorflow as tf
ltn.LAYERS = 10
embedding_size = 10
ltn.BIAS_factor = -1e-5
ltn.set_universal_aggreg("hmean")
closed_pa = pd.read_csv(os.getcwd() + "/../gold_standard/closed_pa",
names=["first", "second", "type"])
closed_an = pd.read_csv(os.getcwd() + "/../gold_standard/closed_an",
names=["first", "second", "type"])
#training_pa = closed_pa.sample(100)
#training_an = closed_an.sample(100)
#test_pa = (pd.concat([training_pa,closed_pa]).drop_duplicates(keep=False))
#test_an = (pd.concat([training_an,closed_an]).drop_duplicates(keep=False))
training_pa = pd.read_csv("training_testing/training_pa")
training_an = pd.read_csv("training_testing/training_an")
test_pa = pd.read_csv("training_testing/test_pa")
help="", type=str)
args = parser.parse_args()
biases = [-1e-8, -1e-5, -1e-1]
lr = args.lr
dc = args.dc
ly = args.ly
embedding_size = args.em
bi = biases[args.bi]
iter_epoch = args.iter
ltn.LAYERS = ly
ltn.BIAS_factor = bi
ltn.set_universal_aggreg(args.univ)
folder_name = "ancestors/reasoning_results/"
entities = ["sue", "diana", "john", "edna", "paul", "francis", "john2",
"john3", "john4", "joe", "jennifer", "juliet", "janice",
"joey", "tom", "bonnie", "katie"]
parents = [
("sue", "diana"),
("john", "diana"),
("sue", "bonnie"),
("john", "bonnie"),
("sue", "tom"),
("john", "tom"),
("diana", "katie"),
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('-const', '--const', help="", type=int)
parser.add_argument('-pred', '--pred', help="", type=int)
parser.add_argument('-card', '--card', help="", type=int)
args = parser.parse_args()
pred = args.pred
const = args.const
card = args.card
ltn.LAYERS = 4
ltn.BIAS_factor = -1e-8
ltn.set_universal_aggreg("mean") # The truth value of forall x p(x) is
# interpretable as the percentage of
# element in the range of x that satisties p
embedding_size = 4 # embedding space dimensionality
pred_name = "pred"
const_name = "const"
predicates = list(map(lambda x: pred_name + str(x), range(1, 100)))
constants = list(map(lambda x: const_name + str(x), range(1, 100)))
for l in constants[:const]:
print(l)
ltnw.constant(l,
min_value=[0.] * embedding_size,