import flor
with flor.Experiment('plate_demo') as ex:
ex.groundClient('ground')
ones = ex.literal([1, 2, 3], "ones")
ones.forEach()
tens = ex.literal([10, 100], "tens")
tens.forEach()
@flor.func
def multiply(x, y):
z = x * y
print(z)
return z
doMultiply = ex.action(multiply, [ones, tens])
product = ex.artifact('product.txt', doMultiply)
product.parallelPull()
# product.plot()
#!/usr/bin/env python3
import flor
import numpy as np
with flor.Experiment("twitter_demo") as ex:
tweets = ex.artifact('tweets.csv', 'tweets_loc')
frac = ex.literal(0.75, 'frac')
split_seed = ex.literal(42, 'seed')
from split import split
do_split = ex.action(split, [tweets, frac, split_seed])
training_tweets = ex.artifact('training_tweets.pkl', 'tweet_df', do_split)
testing_tweets = ex.artifact('testing_tweets.pkl', 'test_df', do_split)
alpha = ex.literalForEach(np.linspace(0.2, 1.0, 2).tolist(), 'alpha')
from train_model import train
do_train = ex.action(train, [training_tweets, alpha])
model = ex.artifact('model.pkl', 'intermediary', do_train)
from test_model2 import test
do_test = ex.action(test, [model, testing_tweets])
model_accuracy = ex.literal(name='model_accuracy', parent=do_test)
model_accuracy.pull()
import flor
@flor.func
def action(literal, artifact, **kwargs):
with open(artifact, 'w') as f:
f.write('Hello\n')
with flor.Experiment('code') as ex:
lit = ex.literal(v=0, name='literal')
do_action = ex.action(action, [lit, ])
art = ex.artifact('artifact', 'artifact', do_action)
art.plot()
#!/usr/bin/env python3
import flor
import numpy as np
ex = flor.Experiment("twitter_demo")
ex.groundClient('ground')
tweets = ex.artifact('tweets.csv')
frac = ex.literal(0.75, 'frac')
split_seed = ex.literal(42, 'split_seed')
from split import split
do_split = ex.action(split, [tweets, frac, split_seed])
training_tweets = ex.artifact('training_tweets.pkl', do_split)
testing_tweets = ex.artifact('testing_tweets.pkl', do_split)
alpha = ex.literal(np.linspace(0.0, 1.0, 8).tolist(), 'alpha')
alpha.forEach()
from train_model import train
do_train = ex.action(train, [training_tweets, alpha])
model = ex.artifact('model.pkl', do_train)
from test_model import test
do_test = ex.action(test, [model, testing_tweets])
model_accuracy = ex.artifact('model_accuracy.txt', do_test)
#!/usr/bin/env python3
import flor
with flor.Experiment('stream') as ex:
@flor.func
def repeat(data, order, dest, **kwargs):
with open(data, 'r') as f:
with open(dest, 'w') as o:
for line in f:
o.write("{}: {}\n".format(order, line))
o.write("{}: {}\n".format(order, line))
order = ex.literalForEach(["first", "second"], "order")
data = ex.artifact('data.txt', 'data')
do_repeat = ex.action(repeat, [data, order])
dest = ex.artifact('dest.txt', 'dest', do_repeat)
dest.plot()
dest.pull()
#!/usr/bin/env python3
import flor
ex = flor.Experiment('twitter')
ex.groundClient('ground')
training_tweets = ex.artifact('training_tweets.csv')
from clean import clean
do_tr_clean = ex.action(clean, [training_tweets])
clean_training_tweets = ex.artifact('clean_training_tweets.pkl', do_tr_clean)
from train_model import train
do_train = ex.action(train, [clean_training_tweets, 0.0001])
intermediary = ex.artifact('intermediary.pkl', do_train)
testing_tweets = ex.artifact('testing_tweets.csv')
do_te_clean = ex.action(clean, [testing_tweets])
clean_testing_tweets = ex.artifact('clean_testing_tweets.pkl', do_te_clean)
from test_model import test
do_test = ex.action(test, [intermediary, clean_testing_tweets])
model_accuracy = ex.artifact('model_accuracy.txt', do_test)
model_accuracy.pull()
model_accuracy.plot()
import flor
with flor.Experiment('coarse') as ex:
ex.groundClient('ground')
@flor.func
def run_existing_pipeline(path_to_data, kernel):
import math
import numpy as np
import pandas as pd
from sklearn import metrics
from sklearn.svm import SVR
from sklearn.model_selection import train_test_split
def manhattan_distance(x1, y1, x2, y2):
return abs(x1 - x2) + abs(y1 - y2)
def roundtime(tstring):
hours, mins, secs = tstring.split(':')
if int(mins) >= 30:
if hours == '23':
return '00'
else:
return str(int(hours) + 1)
else:
return hours
def weekday(start):
from datetime import datetime
fmt = '%Y-%m-%d %H:%M:%S'
@flor.func
def instantiate_agent(**kwargs):
try:
ray.get([])
except:
ray.init()
register_env("my_env", lambda ec: pong_py.PongJSEnv())
trainer = ppo.PPOAgent(env="my_env", config={"env_config": {}})
return trainer
@flor.func
def write_literal(literal, artifact_path, **kwargs):
with open(artifact_path, 'wb') as f:
cloudpickle.dump(literal, f)
with flor.Experiment('nonserial_literal') as ex:
do_instantiate_agent = ex.action(instantiate_agent)
trainer = ex.literal(name="literal", parent=do_instantiate_agent)
do_write_literal = ex.action(write_literal, [
trainer,
])
serialized_trainer = ex.artifact("trainer.pkl", 'artifact_path',
do_write_literal)
serialized_trainer.plot()
serialized_trainer.pull()
#!/usr/bin/env python3
import flor
ex = flor.Experiment('plate_demo')
ex.groundClient('ground')
ones = ex.literal([1, 2, 3], "ones")
ones.forEach()
tens = ex.literal([10, 100], "tens")
tens.forEach()
@flor.func
def multiply(x, y):
z = x*y
print(z)
return z
doMultiply = ex.action(multiply, [ones, tens])
product = ex.artifact('product.txt', doMultiply)
product.plot()
product.pull()
