def main():
parser = arguments.parser(
'Feed Unbound zone information via the control interface')
factory = arguments.Arguments(parser)
try:
factory.read_env()
factory.read_argv()
factory.read_stdin()
except Exception as err:
print(err)
sys.exit(1)
registry = prometheus_client.core.REGISTRY
control = factory.control_client()
provider = factory.provider()
spoon = feeder.Feeder(control, provider)
log_level = logging.INFO
if factory.quiet:
log_level = logging.WARNING
elif factory.verbose:
log_level = logging.DEBUG
logging.basicConfig(format='%(message)s', level=log_level)
provider.enable_logger()
spoon.enable_logger()
if factory.metrics:
spoon.enable_metrics(registry)
prometheus_client.start_http_server(factory.metrics_port,
factory.metrics_bind, registry)
schedule_job(spoon, factory.zone_type, factory.day, factory.time,
factory.instant)
import numpy as np
import random
import arguments
# arguments
args = arguments.parser()
lr = args.lr # learning rate
y = args.y # discount factor
e = args.e # e-greedy factor
n_episodes = args.r # episode number (round)
n_steps = args.s # step number
print("> Setting:", args)
class Agent:
def __init__(self, n_actions, world_size):
self.n_actions = n_actions
self.Q_table = np.zeros([world_size, self.n_actions])
self.size = int(np.sqrt(world_size))
self.lr = lr
self.y = y
self._e0 = e
self.e = self._e0
self._noise = 1.0
self.noise = self._noise
def _flatten(self, cord):
return cord[0] * self.size + cord[1]
matplotlib.use('Agg')
import matplotlib.pyplot as plt
from matplotlib import style
style.use('ggplot')
from gym import spaces
import sys
import os
from arguments import parser, print_args
import pdb
from functions import OUNoise, NormalizedEnv, Memory, Critic, Actor, DDPGagent, make_env, graph_reward, penalties, velocity, shape_rew, step_jor, reward_first, get_observation, shape_reward_s, step_s
from functions import MyProstheticsEnv
import time
cuda = True if torch.cuda.is_available() else False
args = parser()
print_args(args)
#env = make_env(test=args.mode_test,render=args.render_environment)
env = MyProstheticsEnv(integrator_accuracy=1e-4)
env.seed()
#env = NormalizedEnv(env)
obs_size = np.asarray(env.observation_space.shape).prod()
#action_size = np.asarray(action_space.shape).prod()
agent = DDPGagent(env)
noise = OUNoise(env.action_space)
batch_size = args.minibatch_size
rewards = []
avg_rewards = []
best_reward = -10000
from sklearn.metrics import roc_auc_score
from sklearn.metrics import mean_squared_error
from scipy import stats
import torch
import torch.nn as nn
from torch.utils.data import DataLoader
import arguments
import utils
from dataset import MolDataset
from dataset import DTISampler
from dataset import tensor_collate_fn
import model
random.seed(0)
args = arguments.parser(sys.argv)
print(args)
# Read labels
with open(args.filename) as f:
lines = f.readlines()
lines = [l.split() for l in lines]
id_to_y = {l[0]: float(l[1]) for l in lines}
with open(args.key_dir + "/test_keys.pkl", "rb") as f:
test_keys = pickle.load(f)
# Model
cmd = utils.set_cuda_visible_device(args.ngpu)
os.environ["CUDA_VISIBLE_DEVICES"] = cmd[:-1]
if args.potential == "morse":
def test_no_dir(capsys):
with pytest.raises(SystemExit):
arguments.parser([])
out, err = capsys.readouterr()
assert "error: the following arguments are required: -d/--directory" in err
def test_bitrate_string(capsys):
with pytest.raises(SystemExit):
arguments.parser(["-d", "~/Videos/", "-b", "notanint"])
out, err = capsys.readouterr()
assert "error: argument -b/--bitrate: invalid int value: 'notanint'" in err
def test_bitrate_set():
args = arguments.parser(["-d", "~/Videos/", "-b", "{}".format(int(DEFAULT_BITRATE / 2))])
assert args.bitrate == int(DEFAULT_BITRATE / 2)
def test_bitrate_default():
args = arguments.parser(["-d", "~/Videos/"])
assert args.bitrate == DEFAULT_BITRATE
def test_codec_wrong_choice(capsys):
with pytest.raises(SystemExit):
arguments.parser(["-d", "~/Videos/", "-c", "notacodec"])
out, err = capsys.readouterr()
assert "error: argument -c/--codec: invalid choice: 'notacodec' (choose from" in err
def test_codec_h264():
args = arguments.parser(["-d", "~/Videos/", "-c", "h264"])
assert args.videocodec == "h264"
def test_codec_default():
args = arguments.parser(["-d", "~/Videos/"])
assert args.videocodec == "hevc"
def test_single_dir():
args = arguments.parser(["-d", "~/Videos/"])
assert args.paths == ["~/Videos/"]
def main():
args = parser()
print("=" * 20)
print("Experiment parameters:")
for arg in vars(args):
print(arg, getattr(args, arg))
print("=" * 20)
src_repr_path = os.path.join(args.output_path,
f"{args.src_repr_model}_{args.src_name}.csv")
dst_repr_path = os.path.join(args.output_path,
f"{args.dst_repr_model}_{args.dst_name}.csv")
if args.src_repr_model == args.dst_repr_model:
if args.src_repr_model == "BOW":
print("Generating BOW representations for source and destination")
BOW.generate_BOW_vectors_two_taxonomies(args.src_problems_path,
args.dst_problems_path,
src_repr_path,
dst_repr_path)
results = evaluate.evaluate(src_repr_path, dst_repr_path,
args.labels_path, args.recall_at_5,
args.mrr)
elif args.src_repr_model == "TFIDF":
print(
"Generating TFIDF representations for source and destination")
TFIDF.generate_TFIDF_vectors_two_taxonomies(
args.src_problems_path, args.dst_problems_path, src_repr_path,
dst_repr_path)
results = evaluate.evaluate(src_repr_path, dst_repr_path,
args.labels_path, args.recall_at_5,
args.mrr)
elif args.src_repr_model == "content2vec":
print(
"Generating content2vec representations for source and destination"
)
content2vec.generate_content2vec_vectors_two_taxonomies(
args.src_problems_path, args.dst_problems_path, src_repr_path,
dst_repr_path, args.content2vec_size, args.content2vec_window,
args.content2vec_min_count)
results = evaluate.evaluate(src_repr_path, dst_repr_path,
args.labels_path, args.recall_at_5,
args.mrr)
elif args.src_repr_model == "skill2vec":
print(
"Generating skill2vec representations for source and destination"
)
skill2vec.generate_skill2vec_vectors_two_taxonomies(
args.src_sequences_path, args.src_skill2id_path, src_repr_path,
args.src_skill2vec_size, args.src_skill2vec_window,
args.dst_sequences_path, args.dst_skill2id_path, dst_repr_path,
args.dst_skill2vec_size, args.dst_skill2vec_window)
print("Learning skill translation")
training_labels_path, test_labels_path = translate.split_labels(
args.labels_path)
src_translated_path = translate.generate_translated_source_vectors(
src_repr_path, dst_repr_path, training_labels_path,
args.device, args.num_epochs, args.val_ratio,
args.max_val_non_decreasing_epochs, args.verbose)
results = evaluate.evaluate(src_translated_path, dst_repr_path,
test_labels_path, args.recall_at_5,
args.mrr)
elif args.src_repr_model == "content2vec_skill2vec":
src_content2vec_path = os.path.join(
args.output_path, f"content2vec_{args.src_name}.csv")
src_skill2vec_path = os.path.join(
args.output_path, f"skill2vec_{args.src_name}.csv")
dst_content2vec_path = os.path.join(
args.output_path, f"content2vec_{args.dst_name}.csv")
dst_skill2vec_path = os.path.join(
args.output_path, f"skill2vec_{args.dst_name}.csv")
print(
"Generating content2vec representations for source and destination"
)
content2vec.generate_content2vec_vectors_two_taxonomies(
args.src_problems_path, args.dst_problems_path,
src_content2vec_path, dst_content2vec_path,
args.content2vec_size, args.content2vec_window,
args.content2vec_min_count)
print(
"Generating skill2vec representations for source and destination"
)
skill2vec.generate_skill2vec_vectors_two_taxonomies(
args.src_sequences_path, args.src_skill2id_path,
src_skill2vec_path, args.src_skill2vec_size,
args.src_skill2vec_window, args.dst_sequences_path,
args.dst_skill2id_path, dst_skill2vec_path,
args.dst_skill2vec_size, args.dst_skill2vec_window)
print("Learning skill translation")
training_labels_path, test_labels_path = translate.split_labels(
args.labels_path)
src_translated_path = translate.generate_translated_source_vectors(
src_skill2vec_path, dst_skill2vec_path, training_labels_path,
args.device, args.num_epochs, args.val_ratio,
args.max_val_non_decreasing_epochs, args.verbose)
print(
"Combining content2vec and skill2vec representations for source and destination"
)
content2vec_skill2vec.combine_content2vec_and_skill2vec_two_taxonomies(
src_content2vec_path, src_translated_path, src_repr_path,
dst_content2vec_path, dst_skill2vec_path, dst_repr_path)
results = evaluate.evaluate(src_repr_path, dst_repr_path,
test_labels_path, args.recall_at_5,
args.mrr)
elif args.src_repr_model == "TAMF":
src_content2vec_path = os.path.join(
args.output_path, f"content2vec_{args.src_name}.csv")
dst_content2vec_path = os.path.join(
args.output_path, f"content2vec_{args.dst_name}.csv")
print(
"Generating content2vec representations for source and destination"
)
content2vec.generate_content2vec_vectors_two_taxonomies(
args.src_problems_path, args.dst_problems_path,
src_content2vec_path, dst_content2vec_path,
args.content2vec_size, args.content2vec_window,
args.content2vec_min_count)
print("Generating TAMF representations for source and destination")
TAMF.generate_TAMF_vectors_two_taxonomies(
args.src_sequences_path, args.src_skill2id_path,
src_content2vec_path, src_repr_path, args.src_TAMF_k,
args.src_TAMF_lambda, args.dst_sequences_path,
args.dst_skill2id_path, dst_content2vec_path, dst_repr_path,
args.dst_TAMF_k, args.dst_TAMF_lambda)
print("Learning skill translation")
training_labels_path, test_labels_path = translate.split_labels(
args.labels_path)
src_translated_path = translate.generate_translated_source_vectors(
src_repr_path, dst_repr_path, training_labels_path,
args.device, args.num_epochs, args.val_ratio,
args.max_val_non_decreasing_epochs, args.verbose)
results = evaluate.evaluate(src_translated_path, dst_repr_path,
test_labels_path, args.recall_at_5,
args.mrr)
elif args.src_repr_model != args.dst_repr_model:
## Source
if args.src_repr_model == "BOW":
print("Generating BOW representations for source")
BOW.generate_BOW_vectors_one_taxonomy(args.src_problems_path,
src_repr_path)
elif args.src_repr_model == "TFIDF":
print("Generating TFIDF representations for source")
TFIDF.generate_TFIDF_vectors_one_taxonomy(args.src_problems_path,
src_repr_path)
elif args.src_repr_model == "content2vec":
print("Generating content2vec representations for source")
content2vec.generate_content2vec_vectors_one_taxonomy(
args.src_problems_path, src_repr_path,
args.src_content2vec_size, args.src_content2vec_window,
args.src_content2vec_min_count)
elif args.src_repr_model == "skill2vec":
print("Generating skill2vec representations for source")
skill2vec.generate_skill2vec_vectors_one_taxonomy(
args.src_sequences_path, args.src_skill2id_path, src_repr_path,
args.src_skill2vec_size, args.src_skill2vec_window)
elif args.src_repr_model == "content2vec_skill2vec":
src_content2vec_path = os.path.join(
args.output_path, f"content2vec_{args.src_name}.csv")
src_skill2vec_path = os.path.join(
args.output_path, f"skill2vec_{args.src_name}.csv")
print("Generating content2vec representations for source")
content2vec.generate_content2vec_vectors_one_taxonomy(
args.src_problems_path, src_content2vec_path,
args.src_content2vec_size, args.src_content2vec_window,
args.src_content2vec_min_count)
print("Generating skill2vec representations for source")
skill2vec.generate_skill2vec_vectors_one_taxonomy(
args.src_sequences_path, args.src_skill2id_path,
src_skill2vec_path, args.src_skill2vec_size,
args.src_skill2vec_window)
print(
"Combining content2vec and skill2vec representations for source"
)
content2vec_skill2vec.combine_content2vec_and_skill2vec_one_taxonomy(
src_content2vec_path, src_skill2vec_path, src_repr_path)
elif args.src_repr_model == "TAMF":
src_content2vec_path = os.path.join(
args.output_path, f"content2vec_{args.src_name}.csv")
print("Generating content2vec representations for source")
content2vec.generate_content2vec_vectors_one_taxonomy(
args.src_problems_path, src_content2vec_path,
args.src_content2vec_size, args.src_content2vec_window,
args.src_content2vec_min_count)
print("Generating TAMF representations for source")
TAMF.generate_TAMF_vectors_one_taxonomy(args.src_sequences_path,
args.src_skill2id_path,
src_content2vec_path,
src_repr_path,
args.src_TAMF_k,
args.src_TAMF_lambda)
## Destination
if args.dst_repr_model == "BOW":
print("Generating BOW representations for destination")
BOW.generate_BOW_vectors_one_taxonomy(args.dst_problems_path,
dst_repr_path)
elif args.dst_repr_model == "TFIDF":
print("Generating TFIDF representations for destination")
TFIDF.generate_TFIDF_vectors_one_taxonomy(args.dst_problems_path,
dst_repr_path)
elif args.dst_repr_model == "content2vec":
print("Generating content2vec representations for destination")
content2vec.generate_content2vec_vectors_one_taxonomy(
args.dst_problems_path, dst_repr_path,
args.dst_content2vec_size, args.dst_content2vec_window,
args.dst_content2vec_min_count)
elif args.dst_repr_model == "skill2vec":
print("Generating skill2vec representations for destination")
skill2vec.generate_skill2vec_vectors_one_taxonomy(
args.dst_sequences_path, args.dst_skill2id_path, dst_repr_path,
args.dst_skill2vec_size, args.dst_skill2vec_window)
elif args.dst_repr_model == "content2vec_skill2vec":
dst_content2vec_path = os.path.join(
args.output_path, f"content2vec_{args.dst_name}.csv")
dst_skill2vec_path = os.path.join(
args.output_path, f"skill2vec_{args.dst_name}.csv")
print("Generating content2vec representations for destination")
content2vec.generate_content2vec_vectors_one_taxonomy(
args.dst_problems_path, dst_content2vec_path,
args.dst_content2vec_size, args.dst_content2vec_window,
args.dst_content2vec_min_count)
print("Generating skill2vec representations for destination")
skill2vec.generate_skill2vec_vectors_one_taxonomy(
args.dst_sequences_path, args.dst_skill2id_path,
dst_skill2vec_path, args.dst_skill2vec_size,
args.dst_skill2vec_window)
print(
"Combining content2vec and skill2vec representations for destination"
)
content2vec_skill2vec.combine_content2vec_and_skill2vec_one_taxonomy(
dst_content2vec_path, dst_skill2vec_path, dst_repr_path)
elif args.dst_repr_model == "TAMF":
dst_content2vec_path = os.path.join(
args.output_path, f"content2vec_{args.dst_name}.csv")
print("Generating content2vec representations for destination")
content2vec.generate_content2vec_vectors_one_taxonomy(
args.dst_problems_path, dst_content2vec_path,
args.dst_content2vec_size, args.dst_content2vec_window,
args.dst_content2vec_min_count)
print("Generating TAMF representations for destination")
TAMF.generate_TAMF_vectors_one_taxonomy(args.dst_sequences_path,
args.dst_skill2id_path,
dst_content2vec_path,
dst_repr_path,
args.dst_TAMF_k,
args.dst_TAMF_lambda)
print("Learning skill translation")
training_labels_path, test_labels_path = translate.split_labels(
args.labels_path)
src_translated_path = translate.generate_translated_source_vectors(
src_repr_path, dst_repr_path, training_labels_path, args.device,
args.num_epochs, args.val_ratio,
args.max_val_non_decreasing_epochs, args.verbose)
results = evaluate.evaluate(src_translated_path, dst_repr_path,
test_labels_path, args.recall_at_5,
args.mrr)
print("=" * 20)
print(f"results {results}")
json.dump(results, open(os.path.join(args.output_path, "results.json"),
"w"))
