Powered by Aim
| Drop a star to support Aim ⭐ |
Join Aim discord community
|
Run beautiful UI on top of your MLflow logs and get powerful run comparison features.
aimlflow helps to explore various types of metadata tracked during the training with MLFLow, including:
- hyper-parameters
- metrics
- images
- audio
- text
More about Aim: https://github.com/aimhubio/aim
More about MLFLow: https://github.com/mlflow/mlflow
Follow the steps below to set up aimlflow.
- Install aimlflow on your training environment:
pip install aim-mlflow
- Run live time convertor to sync MLFlow logs with Aim:
aimlflow sync --mlflow-tracking-uri={mlflow_uri} --aim-repo={aim_repo_path}
- Run the Aim UI:
aim up --repo={aim_repo_path}
- Powerful pythonic search to select the runs you want to analyze.
- Group metrics by hyperparameters to analyze hyperparameters’ influence on run performance.
- Select multiple metrics and analyze them side by side.
- Aggregate metrics by std.dev, std.err, conf.interval.
- Align x axis by any other metric.
- Scatter plots to learn correlations and trends.
- High dimensional data visualization via parallel coordinate plot.
🎇 Read the article: Exploring MLflow experiments with a powerful UI
🔍 Read the article: How to integrate aimlflow with your remote MLflow
📊 Read the article: Aim and MLflow — Choosing Experiment Tracker for Zero-Shot Cross-Lingual Transfer
