Skip to content
DAGsHub client libraries
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Build Status pypi

DAGsHub Python client libraries

Use DAGsHub to create reproducible versions of your data science research project, allow others to understand your project, and to contribute back to it.

DAGsHub is built firmly around open, standard formats for your project. In particular:

  • git
  • DVC
  • Standard data formats like YAML, JSON, CSV

Therefore, you can work with DAGsHub regardless of your chosen programming language or frameworks.

This client library is meant to help you get started quickly in Python, but it's purely optional - the data formats are very simple and you can choose to work with them directly.

You can learn more by completing our short tutorial or reading the docs


pip install dagshub

Basic Usage

from dagshub import dagshub_logger, DAGsHubLogger

# As a context manager:
with dagshub_logger() as logger:
    # Metrics:
    logger.log_metrics(loss=3.14, step_num=1)
    # OR:
    logger.log_metrics({'val_loss': 6.28}, step_num=2)
    # Hyperparameters:
    # OR:
    logger.log_hyperparams({'optimizer': 'sgd'})

# As a normal Python object:
logger = DAGsHubLogger()
logger.log_metrics(batches_per_second=100, step_num=42)
# ...

Integrations with ML frameworks

The basic DAGsHub logger is just plain Python, and requires no specific framework.

However, for convenience, we include some integrations with common ML frameworks, which can just work right out of the box, without having to write any logging code on your own:

Made with 🐶 by DAGsHub.

You can’t perform that action at this time.