Implementation of a simple persistent value and function cache.
Project description
Copyright 2017 Jeff Ward
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
This is an extremely basic implementation of a value and function cache. There are better tools out there I am sure. It is in initial development so there are likely major bugs. I may update this as I use it more, but you are safest expecting no further updates.
The intent behind this package is to provide something that in one line decorates a function call to, as transparently as possible, cache its results. The use case I built this around was searching hyperparameters. By wrapping your model call in this cache you can quickly implement either random search, or even MCTS, where previously generated results are automatically re-loaded without putting custom caching logic into your code. Additionally, if you are performing grid searches with parameter values from a-c and later realizae you wanted to search from a-m then using this function cache allows you to re-test and automatically re-load existing results while only calculating new results.
A simple example:
import easycache as ec import time def myExpensiveFunction(cost, extraCost=0.0): time.sleep(cost) time.sleep(extraCost) print("This was expensive!") return cost def gridSearchExample(paramA, paramB): """ Simulates a grid search over parameter values :param paramA: list of values :param paramB: list of values :return: """ cachedEF = ec.cacheFunction(myExpensiveFunction, 'example.pkl', name="myExpensiveFunction") bestValue = None bestA = None bestB = None for a in paramA: for b in paramB: newValue = cachedEF(a+b) if bestValue is None or newValue > bestValue: bestValue = newValue bestA = a bestB = b return bestValue, bestA, bestB print("Best Value={} using A={} and B={}".format(*gridSearchExample([.1, .2], [.01, .02]))) #Oh, we REALLY should have searched more! This should only run the 2 new tests print("Best Value={} using A={} and B={}".format(*gridSearchExample([.1, .2], [.01, .02, .003])))
Of course you can do other things:
def runClass(cache): print("Cached Value={}".format(cache.someValue)) #Assignment will flush the cache. #If this is a complex object you will need to manually flush if internal state changes cache.someValue = 12 print("Cached Value={}".format(cache.someValue)) print("peek={}".format(cache.myExpensiveFunction(.5, mode="cache_peek"))) cache.myExpensiveFunction(.5) def classExample(): cache = ec.EasyCache("exampleClass.pkl") cache.clearCache() cache.cacheFunction(myExpensiveFunction, name="myExpensiveFunction") cache.cacheProperty("someValue", initialValue=42) runClass(cache) runClass(cache) cache.myExpensiveFunction(.5,mode="force_run") #need 'mode' for a parameter? change it cache.modeArg='mode2' cache.myExpensiveFunction(.5,mode2="clear_clear") cache.myExpensiveFunction(.5) del cache.someValue del cache.myExpensiveFunction classExample()
And you can ignore parameters for the purpose of caching:
def myComplexExpensiveFunction(cost, uglyState, moreUglyState=None): time.sleep(cost) print("This was expensive and complex!") return cost def ignoreParametersExample(): cachedCEF = ec.cacheFunction(myComplexExpensiveFunction, 'example.pkl', name="myComplexExpensiveFunction", ignoreArgs=(1,), ignorekwArgs=('moreUglyState')) cachedCEF(.5, 'ignored for caching purposes', moreUglyState='also ignored for caching purposes') cachedCEF(.5, 'still cached', moreUglyState='still cached') cachedCEF(.1, 'this wasn\'t cached', moreUglyState='because the arg changed') ignoreParametersExample()
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file easycache-0.1.5.tar.gz
.
File metadata
- Download URL: easycache-0.1.5.tar.gz
- Upload date:
- Size: 5.0 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 832e88f1df96908f6eb89181ba78758bec876e27aa815c0042dc8bc78a0881b9 |
|
MD5 | 0d068c321cf81ab500e4744a4313985b |
|
BLAKE2b-256 | f5a5ca0a44b668910c22c6f3bccaa3b442616a5e9f4cb6564a8d22dc6c28e8ec |
File details
Details for the file easycache-0.1.5-py2.py3-none-any.whl
.
File metadata
- Download URL: easycache-0.1.5-py2.py3-none-any.whl
- Upload date:
- Size: 7.9 kB
- Tags: Python 2, Python 3
- Uploaded using Trusted Publishing? No
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 2c112d109ba3265c3714d6848168337f9a29e8ea45d5df3e16ed0a6130380a52 |
|
MD5 | 37f8ce03886989520cabdfbf8a343b5a |
|
BLAKE2b-256 | 6bee88051ba93c7d107d1cb20f3ce69bfec112869344222d58d82538ee1b57ff |