Skip to content

2.3.0
Compare
Choose a tag to compare
@HydroRobot HydroRobot released this 11 Jun 07:27
· 635 commits to master since this release
2.3.0
2fa129e

Features

  • Python Sagemaker Integration - use Hydrosphere to monitor your machine learning models deployed with Sagemaker. Check hydro-serving-integrations to learn more

  • Python SDK introduces Predictors - object made for seamless conversion between your NumPy or Pandas data representation intro proto messages used by Hydrosphere. Sample code:

# Find application deployed at your Hydrosphere cluster
app = Application.find_by_name(cluster, "my-model")  
predictor = app.predictor()

# Inference your data without converting it to proto objects
df = pd.read_csv("path/to/data.csv")
for row in df.itertuples(index=False):
    predictor.predict(row._as_dict())

Bug Fixes & Other

  • Fixed error when docker push failed silently Hydrospheredata/hydro-serving-manager#34

  • hydrosdk model.read_yaml now does not ignore install_command, training-data and metadata fields

  • Fixed UI bug due to which you could not return to the previous page

  • Fixed bug when UI sometimes does not handle ModelUpdate event from SSE

Assets 3
Loading