forked from plai-group/machine_learning_helpers
-
Notifications
You must be signed in to change notification settings - Fork 1
/
parallel.py
80 lines (66 loc) · 2.75 KB
/
parallel.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
from __future__ import annotations
import contextlib
import multiprocessing
import time
import joblib
import numpy as np
import pandas as pd
from joblib import Parallel, delayed
from sklearn.model_selection import GroupKFold
from tqdm.auto import tqdm
@contextlib.contextmanager
def tqdm_joblib(tqdm_object):
# from https://stackoverflow.com/questions/24983493/tracking-progress-of-joblib-parallel-execution/49950707
"""Context manager to patch joblib to report into tqdm progress bar given as argument"""
class TqdmBatchCompletionCallback(joblib.parallel.BatchCompletionCallBack):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
def __call__(self, *args, **kwargs):
tqdm_object.update(n=self.batch_size)
return super().__call__(*args, **kwargs)
old_batch_callback = joblib.parallel.BatchCompletionCallBack
joblib.parallel.BatchCompletionCallBack = TqdmBatchCompletionCallback
try:
yield tqdm_object
finally:
joblib.parallel.BatchCompletionCallBack = old_batch_callback
tqdm_object.close()
def pmap(f, arr, n_jobs=-1, disable_tqdm=False, **kwargs):
arr = list(arr) # convert generators to list so tqdm works
with tqdm_joblib(tqdm(total=len(arr), disable=disable_tqdm)) as progress_bar:
return Parallel(n_jobs=n_jobs, **kwargs)(delayed(f)(i) for i in arr)
def pmap_df(f, df, n_chunks=100, groups=None, axis=0, **kwargs):
# https://towardsdatascience.com/make-your-own-super-pandas-using-multiproc-1c04f41944a1
if groups:
n_chunks = min(n_chunks, df[groups].nunique())
group_kfold = GroupKFold(n_splits=n_chunks)
df_split = [df.iloc[test_index] for _, test_index in group_kfold.split(df, groups=df[groups])]
else:
df_split = np.array_split(df, n_chunks)
df = pd.concat(pmap(f, df_split, **kwargs), axis=axis)
return df
# For long running jupyter cells
def run_async(func):
"""
# example
# @run_async
# def long_run(idx, val='cat'):
# for i in range(idx):
# print(i)
# time.sleep(1)
# return val
"""
def func_with_queue(queue, *args, **kwargs):
print(f'Running function {func.__name__}{args} {kwargs} ... ')
start_time = time.perf_counter()
result = func(*args, **kwargs)
end_time = time.perf_counter()
total_time = end_time - start_time
queue.put(result)
print(f'Function {func.__name__}{args} {kwargs} Took {total_time:.4f} seconds')
def wrapper(*args, **kwargs):
queue = multiprocessing.Manager().Queue()
process = multiprocessing.Process(target=func_with_queue, args=(queue, *args), kwargs=kwargs)
process.start()
return queue
return wrapper