def predict_by_model_one(data):
global ready_model
if ready_model == False:
ready_model = get_trained_model(model_file)
image = convert_base64_to_grayscale_array(data["image"])
return ready_model.predict([image])
from model import get_trained_model
for key in [('share_count', 100), ('comment_count', 50),
('reaction_count', 50)]:
model = get_trained_model(key[0], key[1], is_logistic=False)
print(key, model.test())
from bottle import post, run, hook, HTTPError, request, response
from textblob import TextBlob, Blobber
from textblob_fr import PatternTagger, PatternAnalyzer
from model import get_trained_model
viral_model = get_trained_model('share_count', 100, is_logistic=True)
comment_model = get_trained_model('comment_count', 50)
share_model = get_trained_model('share_count', 100)
reaction_model = get_trained_model('reaction_count', 50)
def assert_in(val, err):
if not val:
raise HTTPError(status=400, body=err)
def extract_sentiment(phrase):
tb = Blobber(pos_tagger=PatternTagger(), analyzer=PatternAnalyzer())
return (tb(phrase).sentiment[0] / 2) + 0.5
@hook('after_request')
def enable_cors():
"""
You need to add some headers to each request.
Don't use the wildcard '*' for Access-Control-Allow-Origin in production.
"""
response.headers['Access-Control-Allow-Origin'] = '*'
response.headers['Access-Control-Allow-Methods'] = 'POST'
import cv2
from pyzbar.pyzbar import decode
from model import get_trained_model
import os
model = get_trained_model()
if __name__ == '__main__':
for image in os.listdir('results/'):
image = cv2.imread(f'results/{image}')
print(decode(image))
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('envname', type=str)
args = parser.parse_args()
envname = args.envname
with open(os.path.join('expert_data', envname + '.pkl'), 'rb') as f:
expert_data = pickle.load(f)
# data for behavioural cloning
X = expert_data['observations']
Y = expert_data['actions']
from model import get_trained_model
model, metrics = get_trained_model(X, Y)
print(metrics.history['loss'][-1], metrics.history['val_loss'][-1])
env = gym.make(envname)
num_rollouts = 3
max_steps = 1000
render = True
returns = []
for i in range(num_rollouts):
obs = env.reset()
done = False
totalr = 0