Skip to main content

Hyperparameter config file generator.

Project description

tuneconfig Py Versions PyPI version Build Status Documentation Status License: GPL v3

Hyperparameter config file generator and experiment runner.

Quickstart

$ pip install -U tuneconfig

Usage

Config file generator

import pprint

import tuneconfig

# Define parameter formatting function
def format_fn(param):
    fmt = {
        "batch_size": "batch",
        "horizon": "hr",
        "learning_rate": "lr",
        "optimizer": "opt",
        "epochs": None,
        "num_samples": None,
    }
    return fmt.get(param, param)

# Define a configuration template for grid search
config_iterator = tuneconfig.ConfigFactory({
    "batch_size": tuneconfig.grid_search([32, 128]),
    "horizon": 40,
    "learning_rate": tuneconfig.grid_search([0.01, 0.1]),
    "epochs": 1000,
    "optimizer": tuneconfig.grid_search(["Adam", "RMSProp"]),
    "num_samples": 10
    },
    format_fn=format_fn
)

# Iterate over config dicts
for idx, config in enumerate(config_iterator):
    name = config_iterator._trial_id(config)
    print(f"config {idx} ({name}):")
    pprint.pprint(config)
    print()

# Dump config dicts as JSON files
tmp = "/tmp/tuneconfig"
json_config_files = config_iterator.dump(tmp)
print(">> Saved config files:")
pprint.pprint(json_config_files)
config 0 (batch=32/hr=40/lr=0.01/opt=Adam):
{'batch_size': 32,
 'epochs': 1000,
 'horizon': 40,
 'learning_rate': 0.01,
 'num_samples': 10,
 'optimizer': 'Adam'}

config 1 (batch=32/hr=40/lr=0.01/opt=RMSProp):
{'batch_size': 32,
 'epochs': 1000,
 'horizon': 40,
 'learning_rate': 0.01,
 'num_samples': 10,
 'optimizer': 'RMSProp'}

config 2 (batch=32/hr=40/lr=0.1/opt=Adam):
{'batch_size': 32,
 'epochs': 1000,
 'horizon': 40,
 'learning_rate': 0.1,
 'num_samples': 10,
 'optimizer': 'Adam'}

config 3 (batch=32/hr=40/lr=0.1/opt=RMSProp):
{'batch_size': 32,
 'epochs': 1000,
 'horizon': 40,
 'learning_rate': 0.1,
 'num_samples': 10,
 'optimizer': 'RMSProp'}

config 4 (batch=128/hr=40/lr=0.01/opt=Adam):
{'batch_size': 128,
 'epochs': 1000,
 'horizon': 40,
 'learning_rate': 0.01,
 'num_samples': 10,
 'optimizer': 'Adam'}

config 5 (batch=128/hr=40/lr=0.01/opt=RMSProp):
{'batch_size': 128,
 'epochs': 1000,
 'horizon': 40,
 'learning_rate': 0.01,
 'num_samples': 10,
 'optimizer': 'RMSProp'}

config 6 (batch=128/hr=40/lr=0.1/opt=Adam):
{'batch_size': 128,
 'epochs': 1000,
 'horizon': 40,
 'learning_rate': 0.1,
 'num_samples': 10,
 'optimizer': 'Adam'}

config 7 (batch=128/hr=40/lr=0.1/opt=RMSProp):
{'batch_size': 128,
 'epochs': 1000,
 'horizon': 40,
 'learning_rate': 0.1,
 'num_samples': 10,
 'optimizer': 'RMSProp'}

>> Saved config files:
['/tmp/tuneconfig/batch=32/hr=40/lr=0.01/opt=Adam/config.json',
 '/tmp/tuneconfig/batch=32/hr=40/lr=0.01/opt=RMSProp/config.json',
 '/tmp/tuneconfig/batch=32/hr=40/lr=0.1/opt=Adam/config.json',
 '/tmp/tuneconfig/batch=32/hr=40/lr=0.1/opt=RMSProp/config.json',
 '/tmp/tuneconfig/batch=128/hr=40/lr=0.01/opt=Adam/config.json',
 '/tmp/tuneconfig/batch=128/hr=40/lr=0.01/opt=RMSProp/config.json',
 '/tmp/tuneconfig/batch=128/hr=40/lr=0.1/opt=Adam/config.json',
 '/tmp/tuneconfig/batch=128/hr=40/lr=0.1/opt=RMSProp/config.json']

License

Copyright (c) 2020 Thiago Pereira Bueno All Rights Reserved.

tuneconfig is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

tuneconfig is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details.

You should have received a copy of the GNU Lesser General Public License along with tuneconfig. If not, see http://www.gnu.org/licenses/.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

tuneconfig-0.14.1.tar.gz (15.1 kB view hashes)

Uploaded Source

Built Distribution

tuneconfig-0.14.1-py3-none-any.whl (33.4 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page