Say I load a big data file used within a module but only care about something derived from it that requires very little memory. What is the best way to structure a module so that it doesn't keep unwanted data in memory when I load it as a module.
Something like:
# getstuff.py
from importlib_resources import files
data = files('path.to.file').joinpath('big.one').read_text()
stuff_i_care_about = some_complicated_operation(data)
# __init__.py
from .getstuff import stuff_i_care_about
What is the best way to make sure that data is freed? del + gc.collect()? Wrap it in a function? Might it be freed automatically in some versions anyway?
I am trying to pickle a sklearn machine-learning model, and load it in another project. The model is wrapped in pipeline that does feature encoding, scaling etc. The problem starts when i want to use self-written transformers in the pipeline for more advanced tasks.
Let's say I have 2 projects:
train_project: it has the custom transformers in src.feature_extraction.transformers.py
use_project: it has other things in src, or has no src catalog at all
If in "train_project" I save the pipeline with joblib.dump(), and then in "use_project" i load it with joblib.load() it will not find something such as "src.feature_extraction.transformers" and throw exception:
ModuleNotFoundError: No module named 'src.feature_extraction'
I should also add that my intention from the beginning was to simplify usage of the model, so programist can load the model as any other model, pass very simple, human readable features, and all "magic" preprocessing of features for actual model (e.g. gradient boosting) is happening inside.
I thought of creating /dependencies/xxx_model/ catalog in root of both projects, and store all needed classes and functions in there (copy code from "train_project" to "use_project"), so structure of projects is equal and transformers can be loaded. I find this solution extremely inelegant, because it would force the structure of any project where the model would be used.
I thought of just recreating the pipeline and all transformers inside "use_project" and somehow loading fitted values of transformers from "train_project".
The best possible solution would be if dumped file contained all needed info and needed no dependencies, and I am honestly shocked that sklearn.Pipelines seem to not have that possibility - what's the point of fitting a pipeline if i can not load fitted object later? Yes it would work if i used only sklearn classes, and not create custom ones, but non-custom ones do not have all needed functionality.
Example code:
train_project
src.feature_extraction.transformers.py
from sklearn.pipeline import TransformerMixin
class FilterOutBigValuesTransformer(TransformerMixin):
def __init__(self):
pass
def fit(self, X, y=None):
self.biggest_value = X.c1.max()
return self
def transform(self, X):
return X.loc[X.c1 <= self.biggest_value]
train_project
main.py
from sklearn.externals import joblib
from sklearn.preprocessing import MinMaxScaler
from src.feature_extraction.transformers import FilterOutBigValuesTransformer
pipeline = Pipeline([
('filter', FilterOutBigValuesTransformer()),
('encode', MinMaxScaler()),
])
X=load_some_pandas_dataframe()
pipeline.fit(X)
joblib.dump(pipeline, 'path.x')
test_project
main.py
from sklearn.externals import joblib
pipeline = joblib.load('path.x')
The expected result is pipeline loaded correctly with transform method possible to use.
Actual result is exception when loading the file.
I found a pretty straightforward solution. Assuming you are using Jupyter notebooks for training:
Create a .py file where the custom transformer is defined and import it to the Jupyter notebook.
This is the file custom_transformer.py
from sklearn.pipeline import TransformerMixin
class FilterOutBigValuesTransformer(TransformerMixin):
def __init__(self):
pass
def fit(self, X, y=None):
self.biggest_value = X.c1.max()
return self
def transform(self, X):
return X.loc[X.c1 <= self.biggest_value]
Train your model importing this class from the .py file and save it using joblib.
import joblib
from custom_transformer import FilterOutBigValuesTransformer
from sklearn.externals import joblib
from sklearn.preprocessing import MinMaxScaler
pipeline = Pipeline([
('filter', FilterOutBigValuesTransformer()),
('encode', MinMaxScaler()),
])
X=load_some_pandas_dataframe()
pipeline.fit(X)
joblib.dump(pipeline, 'pipeline.pkl')
When loading the .pkl file in a different python script, you will have to import the .py file in order to make it work:
import joblib
from utils import custom_transformer # decided to save it in a utils directory
pipeline = joblib.load('pipeline.pkl')
Apparently this problem raises when you split definitions and saving code part in two different files. So I have found this workaround that has worked for me.
It consists in these steps:
Guess we have your 2 projects/repositories : train_project and use_project
train_project:
On your train_project create a jupyter notebook or .py
On that file lets define every Custom transformer in a class, and import all other tools needed from sklearn to design the pipelines. Then lets write the saving code to pickle just inside the same file.(Don't create an external .py file src.feature_extraction.transformers to define your customtransformers).
Then fit and dumb your pipeline by running that file.
On use_project:
Create a customthings.py file with all the functions and transformers defined inside.
Create another file_where_load.py where you wish load the pickle. Inside, make sure you have imported all the definitions from customthings.py . Ensure that functions and classes have the same name than the ones you've used on train_project.
I hope it works for everyone with same problem
I have created a workaround solution. I do not consider it a complete answer to my question, but non the less it let me move on from my problem.
Conditions for the workaround to work:
I. Pipeline needs to have only 2 kinds of transformers:
sklearn transformers
custom transformers, but with only attributes of types:
number
string
list
dict
or any combination of those e.g. list of dicts with strings and numbers. Generally important thing is that attributes are json serializable.
II. names of pipeline steps need to be unique (even if there is pipeline nesting)
In short model would be stored as a catalog with joblib dumped files, a json file for custom transformers, and a json file with other info about model.
I have created a function that goes through steps of a pipeline and checks __module__ attribute of transformer.
If it finds sklearn in it it then it runs joblib.dump function under a name specified in steps (first element of step tuple), to some selected model catalog.
Otherwise (no sklearn in __module__) it adds __dict__ of transformer to result_dict under a key equal to name specified in steps. At the end I json.dump the result_dict to model catalog under name result_dict.json.
If there is a need to go into some transformer, because e.g. there is a Pipeline inside a pipeline, you can probably run this function recursively by adding some rules to the beginning of the function, but it becomes important to have always unique steps/transformers names even between main pipeline and subpipelines.
If there are other information needed for creation of model pipeline then save them in model_info.json.
Then if you want to load the model for usage:
You need to create (without fitting) the same pipeline in target project. If pipeline creation is somewhat dynamic, and you need information from source project, then load it from model_info.json.
You can copy function used for serialization and:
replace all joblib.dump with joblib.load statements, assign __dict__ from loaded object to __dict__ of object already in pipeline
replace all places where you added __dict__ to result_dict with assignment of appropriate value from result_dict to object __dict__ (remember to load result_dict from file beforehand)
After running this modified function, previously unfitted pipeline should have all transformer attributes that were effect of fitting loaded, and pipeline as a whole ready to predict.
The main things I do not like about this solution is that it needs pipeline code inside target project, and needs all attrs of custom transformers to be json serializable, but I leave it here for other people that stumble on a similar problem, maybe somebody comes up with something better.
I was similarly surprised when I came across the same problem some time ago. Yet there are multiple ways to address this.
Best practice solution:
As others have mentioned, the best practice solution is to move all dependencies of your pipeline into a separate Python package and define that package as a dependency of your model environment.
The environment then has to be recreated whenever the model is deployed. In simple cases this can be done manually e.g. via virtualenv or Poetry. But model stores and versioning frameworks (MLflow being one example) typically provide a way to define the required Python environment (e.g. via conda.yaml). They often can automatically recreate the environment at deployment time.
Solution by putting code into main:
In fact, class and function declearations can be serialized, but only declarations in __main__ actually get serialized. __main__ is the entry point of the script, the file that is run. So if all the custom code and all of its dependencies are in that file, then custom objects can later be loaded in Python environments that do not include the code. This kind of solves the problem, but who wants to have all that code in __main__? (Note that this property also applies to cloudpickle)
Solution by "mainifying":
There is one other way which is to "mainify" the classes or function objects before saving. I came across that same problem some time ago and have written a function that does that. It essentially redefines an existing object's code in __main__. Its application is simple: Pass object to function, then serialize the object, voilà, it can be loaded anywhere. Like so:
# ------ In file1.py: ------
class Foo():
pass
# ------ In file2.py: ------
from file1 import Foo
foo = Foo()
foo = mainify(foo)
import dill
with open('path/file.dill', 'wb') as f
dill.dump(foo, f)
I post the function code below. Note that I have tested this with dill, but I think it should work with pickle as well.
Also note that the original idea is not mine, but came from a blog post that I cannot find right now. I will add the reference/acknowledgement when I find it.
Edit: Blog post by Oege Dijk by which my code was inspired.
def mainify(obj, warn_if_exist=True):
''' If obj is not defined in __main__ then redefine it in main. Allows dill
to serialize custom classes and functions such that they can later be loaded
without them being declared in the load environment.
Parameters
---------
obj : Object to mainify (function or class instance)
warn_if_exist : Bool, default True. Throw exception if function (or class) of
same name as the mainified function (or same name as mainified
object's __class__) was already defined in __main__. If False
don't throw exception and instead use what was defined in
__main__. See Limitations.
Limitations
-----------
Assumes `obj` is either a function or an instance of a class.
'''
if obj.__module__ != '__main__':
import __main__
is_func = True if isinstance(obj, types.FunctionType) else False
# Check if obj with same name is already defined in __main__ (for funcs)
# or if class with same name as obj's class is already defined in __main__.
# If so, simply return the func with same name from __main__ (for funcs)
# or assign the class of same name to obj and return the modified obj
if is_func:
on = obj.__name__
if on in __main__.__dict__.keys():
if warn_if_exist:
raise RuntimeError(f'Function with __name__ `{on}` already defined in __main__')
return __main__.__dict__[on]
else:
ocn = obj.__class__.__name__
if ocn in __main__.__dict__.keys():
if warn_if_exist:
raise RuntimeError(f'Class with obj.__class__.__name__ `{ocn}` already defined in __main__')
obj.__class__ = __main__.__dict__[ocn]
return obj
# Get source code and compile
source = inspect.getsource(obj if is_func else obj.__class__)
compiled = compile(source, '<string>', 'exec')
# "declare" in __main__, keeping track which key of __main__ dict is the new one
pre = list(__main__.__dict__.keys())
exec(compiled, __main__.__dict__)
post = list(__main__.__dict__.keys())
new_in_main = list(set(post) - set(pre))[0]
# for function return mainified version, else assign new class to obj and return object
if is_func:
obj = __main__.__dict__[new_in_main]
else:
obj.__class__ = __main__.__dict__[new_in_main]
return obj
Have you tried using cloud pickle?
https://github.com/cloudpipe/cloudpickle
Based on my research it seems that the best solution is to create a Python package that includes your trained pipeline and all files.
Then you can pip install it in the project where you want to use it and import the pipeline with from <package name> import <pipeline name>.
Credit to Ture Friese for mentioning cloudpickle >=2.0.0, but here's an example for your use case.
import cloudpickle
cloudpickle.register_pickle_by_value(FilterOutBigValuesTransformer)
with open('./pipeline.cloudpkl', mode='wb') as file:
pipeline.dump(
obj=Pipe
, file=file
)
register_pickle_by_value() is the key as it will ensure your custom module (src.feature_extraction.transformers) is also included when serializing your primary object (pipeline). However, this is not built for recursive module dependence, e.g. if FilterOutBigValuesTransformer also contains another import statement
Calling the location of the transform.py file with sys.path.append may resolve the issue.
import sys
sys.path.append("src/feature_extraction/transformers")
I am trying to parallelize a function that takes in an object in Python:
In using Pathos, the map function automatically dills the object before distributing it among the processors.
However, it takes ~1 min to dill the object each time, and I need run this function up to 100 times. All in all, it is taking nearly 2 hours to just serialize the object before even running it.
Is there a way to just serialize it once, and use it multiple times?
Thanks very much
The easiest thing to do is to do this manually.
Without an example of your code, I have to make a lot of assumptions and write something pretty vague, so let's take the simplest case.
Assume you're using dill manually, so your existing code looks like this:
obj = function_that_creates_giant_object()
for i in range(zillions):
results.append(pool.apply(func, (dill.dumps(obj),)))
All you have to do is move the dumps out of the loop:
obj = function_that_creates_giant_object()
objpickle = dill.dumps(obj)
for i in range(zillions):
results.append(pool.apply(func, (objpickle,)))
But depending on your actual use, it may be better to just stick a cache in front of dill:
cachedpickle = functools.lru_cache(maxsize=10)(dill.dumps)
obj = function_that_creates_giant_object()
for i in range(zillions):
results.append(pool.apply(wrapped_func, (cachedpickle(obj),))
Of course if you're monkeypatching multiprocessing to use dill in place of pickle, you can just as easy patch it to use this cachedpickle function.
If you're using multiprocess, which is a forked version of multiprocessing that pre-substitutes dill for pickle, it's less obvious how to patch that; you'll need to go through the source and see where it's using dill and get it to use your wrapper. But IIRC, it just does a import dill as pickle somewhere and then uses the same code as (a slightly out-of-date version of multiprocessing), so it isn't all that different.
In fact, you can even write a module that exposes the same interface as pickle and dill:
import functools
import dill
def loads(s):
return dill.loads(s)
#lru_cache(maxsize=10)
def dumps(o):
return dill.dumps(o)
… and just replace the import dill as pickle with import mycachingmodule as pickle.
… or even monkeypatch it after loading with multiprocess.helpers.pickle = mycachingmodule (or whatever the appropriate name is—you're still going to have to find where that relevant import happens in the source of whatever you're using).
And that's about as complicated as it's likely to get.
What I have is a class that inherits from DataFrame, but overrides some behavior for business logic reasons. All is well and good, but I need the ability to import and export them. msgpack appears to be a good choice, but doesn't actually work. (Using the standard msgpack library doesn't even work on regular Dataframes, and the advice there is to use the dataframe msgpack functions.)
class DataFrameWrap(pandas.DataFrame):
pass
df = DataFrameWrap()
packed_df = df.to_msgpack()
pandas.read_msgpack(packed_df)
This results in the error
File "C:\Users\REDACTED\PROJECT_NAME\lib\site-packages\pandas\io\packers.py", line 627, in decode
return globals()[obj[u'klass']](BlockManager(blocks, axes))
KeyError: u'DataFrameWrap'
when it reaches the read_msgpack() line. This works if I replace the DataFrameWrap() with a regular DataFrame().
Is there a way to tell pandas where to find the DataFrameWrap class? From reading the code, it looks like if I could inject {"DataFrameWrap": DataFrameWrap} into the globals as seen from this file, it would work, but I'm not sure how to actually do that. There also might be a proper way to do this, but it's not obvious.
Figured it out. As usual, it was much less complicated than I assumed:
from pandas.io import packers
class DataFrameWrap(pandas.DataFrame):
pass
packers.DataFrameWrap = DataFrameWrap
df = DataFrameWrap()
packed_df = df.to_msgpack()
pandas.read_msgpack(packed_df)
I am using JSON to send data from Python to R (note: I'm much more familiar with R than Python). For primitives, the json module works great. For many other Python objects (e.g. numpy arrays) you have to define a custom encoder, like in this stack overflow answer. However, that requires you to pass the encoder as an argument to json.dumps, which doesn't work that well for my case.
I know there are other packages like json_tricks that have much more advanced capabilities for JSON serialization, but since I don't have control over what Python distribution a user has I don't want to rely on any non-default modules for serializing objects to JSON.
I'm wondering if there is a way to use contextlib decorators to define additional ways for serializing JSON objects. Ideally, I'm looking for a way that would allow users to overload some standard function standard_wrapper that I provide to add new methods for their own classes (or types from modules that they load) without requiring them to modify standard_wrapper. Some psuedocode below:
import json
def standard_wrapper(o):
return o
obj = [44,64,13,4,79,2,454,89,0]
json.dumps(obj)
json.dumps(standard_wrapper(obj))
import numpy as np
objnp = np.sort(obj)
json.dumps(objnp) # FAILS
#some_decorator_to_overload_standard_wrapper
# some code
json.dumps(standard_wrapper(objnp)) # HOPEFULLY WORKS
This is essentially function overloading by type---I've seen examples for overloading by arguments in Python, but I don't see how to do it by type.
EDIT I was mixing up decorators with contextlib (which I had only ever seen used a decorator).
It's easy to use singledispatch from functools module to overload a function by type, as shown in this answer to a different post. However, a simpler solution that may fit my needs is to create a dictionary of functions where the keys correspond to the object type.
import numpy
func_dict = {}
a = [2,5,2,9,75,8,36,2,8]
an = numpy.sort(a)
func_dict[type(an)] = lambda x: x.tolist()
func_dict[type(a)] = lambda x: x
import json
json.dumps(func_dict[type(a)](a))
json.dumps(func_dict[type(an)](an))
Adding support for another type is achieved by adding another function to the dictionary.