https://github.com/johnsmithm/htuneml
machine learning experiments monitoring and hyperparameter tuning
https://github.com/johnsmithm/htuneml
experiment-machine-learning experiment-monitor hyperparameter-tuning
Last synced: 3 months ago
JSON representation
machine learning experiments monitoring and hyperparameter tuning
- Host: GitHub
- URL: https://github.com/johnsmithm/htuneml
- Owner: johnsmithm
- Created: 2019-02-14T15:38:36.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2019-02-14T17:00:50.000Z (over 6 years ago)
- Last Synced: 2025-02-26T14:42:42.150Z (4 months ago)
- Topics: experiment-machine-learning, experiment-monitor, hyperparameter-tuning
- Size: 2.93 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# HTUNEML - machine learning experiments monitoring and tuning
**Quickstart:** ``pip install htuneml``. See the "Installing" section for more details.
Project links:
- `PyPI `__
- `Source code `__
- `Issue tracker `__
Examples
--------
See the ``examples/`` `directory `__ in the repository root for usage examples:
- `Tensorflow Mnist `__
- `Keras Mnist `__
- `Pytorch Mnist `__
Requirements
------------
To use all of the functionality of the library, you should have:
* **Python** 2.6, 2.7, or 3.3+ (required)
* **PyAudio** 0.2.11+ (required only if you need to use microphone input, ``Microphone``)
Quick start
------------
Register on website http://registru.ml, copy the api_key:
.. code:: python
import htuneml as ht
job = Job('api_key')
@job.monitor
def train(par1=2,par2=2):
for i in range(par1):
#do training here
job.log({'loss':i*4,'ep':i})
job.setName('l2')
#job.debug()# uncomment and no experiment will be created and no logs sent
train(10, 2)
This will print out something like the following:
::
make experiment
got key experimnet 5c5c8eaacbcfb9146641367a
Also it is possible to sent the parameters from the web app. First on gpu/cpu set the lisener:
.. code:: python
import htuneml as ht
job = Job('api_key')
def train(par1=2,par2=2):
for i in range(par1):
#do training here
job.log({'loss':i*4,'ep':i})
job.sentParams(train)#sent the parameters list to the app
job.waitTask(train)#wait for parameters from app