optdash

Optdash is a third-party dashboard for optuna.

This dashboard is deprecated in favor of optuna/optuna-dashboard.

Key features

• Simple usage
• Interactive visualization
• Minimal dependencies
  • Optdash depends solely on Optuna (and sklearn, which is an optional depedency of optuna).

Quick start

optdash your/db/url

The db url is the same with optuna, whose format follows RFC-1738.

Install

From pypi

pip install optdash

From source

cd optdash/frontend
npm i
npm run build
cd ../../
python -m pip install .

Tutorial

First, create optuna studies.

import optuna

def objective(trial):
     ret = 0
     for j in range(20):
         ret += trial.suggest_float(f"param{j}", 0.0, 1.0)
         trial.report(ret, j)
     return ret

for i in range(10):
     optuna.create_study(
         study_name=f"study{i}", storage="sqlite:///test.db", load_if_exists=True
     ).optimize(objective, n_trials=30)

Next, run optdash from a command prompt.

optdash sqlite:///test.db

Then, you'll see the following message

INFO:root:Serving at http://localhost:8080

Open your web-browser, e.g., Google Chrome, and access to http://localhost:8080.