Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/aistairc/aiaccel
A hyperparameter optimization library for the ABCI.
https://github.com/aistairc/aiaccel
abci hyperparameter-optimization nelder-mead-algorithm
Last synced: 2 days ago
JSON representation
A hyperparameter optimization library for the ABCI.
- Host: GitHub
- URL: https://github.com/aistairc/aiaccel
- Owner: aistairc
- License: mit
- Created: 2022-05-27T09:16:17.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2024-08-23T02:02:15.000Z (3 months ago)
- Last Synced: 2024-08-23T02:43:12.297Z (3 months ago)
- Topics: abci, hyperparameter-optimization, nelder-mead-algorithm
- Language: Python
- Homepage: https://aistairc.github.io/aiaccel/
- Size: 12.7 MB
- Stars: 23
- Watchers: 5
- Forks: 5
- Open Issues: 9
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# aiaccel: an HPO library for ABCI
[](https://github.com/aistairc/aiaccel)
[](https://github.com/aistairc/aiaccel)
[](https://github.com/aistairc/aiaccel/actions/workflows/publish_pages.yaml)
[**日本語 (Japanese)**](https://github.com/aistairc/aiaccel/blob/main/README_JP.md)
A hyperparameter optimization library for [AI Bridging Cloud Infrastructure (ABCI)](https://abci.ai/).
This software solves hyperparameter optimizations related to AI technologies including deep learning and multi-agent simulation.
The software currently supports five optimization algorithms: random search, grid search, sobol sequence, nelder-mead method, and TPE.
# Installation
The software can be installed using `pip`.
~~~bash
> pip install git+https://github.com/aistairc/aiaccel.git
~~~
# Getting started
## Running on a local computer
An example for optimizing a simple function (i.e., sphere function) on a local computer.
0. (Optional) Install [Virtualenv](https://virtualenv.pypa.io/en/latest/) and create a virtual environment.
~~~bash
> python3 -m venv optenv
> source optenv/bin/activate
~~~
1. Install `aiaccel`
~~~bash
> pip install git+https://github.com/aistairc/aiaccel.git
~~~
2. Create a workspace and copy the sphere example on the repository.
~~~bash
> mkdir your_workspace_directory
> cd your_workspace_directory
> git clone https://github.com/aistairc/aiaccel.git
> cp -R ./aiaccel/examples .
> cd examples
> ls
sphere
> cd sphere
> ls
config.yaml user.py
~~~
3. Run the parameter optimization
~~~bash
> aiaccel-start --config config.yaml
~~~
or
~~~bash
> python -m aiaccel.cli.start --config config.yaml
~~~
Tips: You can start after cleaning the workspace directory using `--clean`.
~~~bash
> aiaccel-start --config config.yaml --clean
~~~
4. Wait for the program to finish and check the optimization results.
~~~bash
> ls ./work
abci_output alive hp lock
log result runner state
> cat ./work/result/final_result.result
~~~
5. If you want to change configurations, edit `config.yaml`.
~~~bash
> vi config.yaml
~~~
## Running on ABCI
This tutorial describes how to run examples/sphere on ABCI.
1. First, setup python environment following [the ABCI Users Guide](https://docs.abci.ai/en/python/):
~~~bash
> module load python/3.11/3.11.2
> python3 -m venv optenv
> source optenv/bin/activate
~~~
2. Prepare the workspace by following Steps 1 and 2 in [Running on a local computer](https://github.com/aistairc/aiaccel#Running-on-a-local-computer).
3. Please confirm the configuration file before running master.
```yaml
resource:
type: "abci"
num_workers: 4
```
4. Run on an (interactive) job
~~~bash
> aiaccel-start --config config.yaml
~~~
5. If you want to check the running jobs, please refer the [ABCI User Guide](https://docs.abci.ai/en/job-execution/#show-the-status-of-batch-jobs).
## Others
- Check the progress
~~~bash
> aiaccel-view --config config.yaml
~~~
- Display simple graphs
~~~bash
> aiaccel-plot --config config.yaml
~~~
- Output results to workspace/results.csv.
~~~bash
> aiaccel-report --config config.yaml
~~~
# Acknowledgement
* Part of this software was developed in a project commissioned by the New Energy and Industrial Technology Development Organization (NEDO).
* aiaccel is built with the help of Optuna.