Installation
pip install memo
Documentation
The documentation can be found here.
The quickstart guide is found here.
Usage
Here's an example of utility functions provided by our library.
import numpy as np
from memo import memlist, memfile, grid, time_taken
data = []
@memfile(filepath="results.jsonl")
@memlist(data=data)
@time_taken()
def birthday_experiment(class_size, n_sim):
"""Simulates the birthday paradox. Vectorized = Fast!"""
sims = np.random.randint(1, 365 + 1, (n_sim, class_size))
sort_sims = np.sort(sims, axis=1)
n_uniq = (sort_sims[:, 1:] != sort_sims[:, :-1]).sum(axis = 1) + 1
proba = np.mean(n_uniq != class_size)
return {"est_proba": proba}
for settings in grid(class_size=[5, 10, 20, 30], n_sim=[1000, 1_000_000]):
birthday_experiment(**settings)The decorators memlist and memfile are making sure that the keyword arugments and
dictionary output of the birthday_experiment are logged. The contents of the results.jsonl-file
and the data variable looks like this;
{"class_size": 5, "n_sim": 1000, "est_proba": 0.024, "time_taken": 0.0004899501800537109}
{"class_size": 5, "n_sim": 1000000, "est_proba": 0.027178, "time_taken": 0.19407916069030762}
{"class_size": 10, "n_sim": 1000, "est_proba": 0.104, "time_taken": 0.000598907470703125}
{"class_size": 10, "n_sim": 1000000, "est_proba": 0.117062, "time_taken": 0.3751380443572998}
{"class_size": 20, "n_sim": 1000, "est_proba": 0.415, "time_taken": 0.0009679794311523438}
{"class_size": 20, "n_sim": 1000000, "est_proba": 0.411571, "time_taken": 0.7928380966186523}
{"class_size": 30, "n_sim": 1000, "est_proba": 0.703, "time_taken": 0.0018239021301269531}
{"class_size": 30, "n_sim": 1000000, "est_proba": 0.706033, "time_taken": 1.1375510692596436}
The nice thing about being able to log results to a file or to the web is that
you can also more easily parallize your jobs! For example now you can use the Runner
class to parrallelize the function call with joblib.
import numpy as np
from memo import memlist, memfile, grid, time_taken, Runner
data = []
@memfile(filepath="results.jsonl")
@memlist(data=data)
@time_taken()
def birthday_experiment(class_size, n_sim):
"""Simulates the birthday paradox. Vectorized = Fast!"""
sims = np.random.randint(1, 365 + 1, (n_sim, class_size))
sort_sims = np.sort(sims, axis=1)
n_uniq = (sort_sims[:, 1:] != sort_sims[:, :-1]).sum(axis = 1) + 1
proba = np.mean(n_uniq != class_size)
return {"est_proba": proba}
# declare all the settings to loop over
settings = grid(class_size=range(20, 30), n_sim=[100, 10_000, 1_000_000])
# use a runner to run over all the settings
runner = Runner(backend="threading", n_jobs=-1)
runner.run(func=birthday_experiment, settings=settings, progbar=True)Features
This library also offers decorators to pipe to other sources.
memlistsends the json blobs to a listmemfilesends the json blobs to a filememwebsends the json blobs to a server via http-post requestsmemfuncsends the data to a callable that you supply, likeprintgridgenerates a convenient grid for your experimentsrandom_gridgenerates a randomized grid for your experimentstime_takenalso logs the time the function takes to run
We also offer an option to parallelize function calls using joblib. This
is facilitated with a Runner class which supports multiple backends.
Runner(backend="loky")Runner(backend="threading")Runner(backend="multiprocessing")
Check the API docs here for more information on how these work.