0.4.0

🚀 Features

• Add "Absolute Average Fold Error" and "Balanced Average Precision" to the metrics
  • PR: #101

0.3.0

🚀 Features

• feat: Listings Caching implementation using fsspec dircache
  • PR: #93
• feat: Copy a Zarr archive to the Hub using copy_store for increased efficient
  • PR: #92

🐛 Fixes

• Adding support for MIT license on client
  • PR: #100

📚 Documentation

• Update issue templates
  • PR: #94

0.2.6

🚀 Features

• Add a flow to use Zarr datasets through the Hub
  • PR: #83
• Added the version attribute to the artifact data model
  • PR: #84

🐛 Fixes

• Fix a bug in Dataset.cache() and set a default timeout for the PolarisHubClient
  • PR: #85
• Updating column adapter logic and usage
  • PR: #86
• Fix "Connection reset by peer" error when uploading large-ish Zarr datasets
  • PR: #88
• Fix: Passing wrong Zarr store that caused metadata consolidation error in upload_dataset()
  • PR: #90
• Preventing certain license types on dataset upload to hub
  • PR: #89

📚 Documentation

• Changed the community references in the docs from Portal to Discord
  • PR: #81

0.2.5

🐛 Fixes

• Hotfix: Added option to disable the checksum verification
  • PR: #80

0.2.4

🚀 Features

• feature: polaris fsspec basic read only implementation
  • PR: #72
• Easily featurize the inputs in the Subset class
  • PR: #71
• feature: polaris fsspec write implementation
  • PR: #73
• Add additional features to support SDF to Zarr conversion
  • PR: #77

🐛 Fixes

• Fix default settings for the PolarisHubSettings
  • PR: #78

🧪 Tests

• Easily featurize the inputs in the Subset class
  • PR: #71

📚 Documentation

• Improve the documentation of the Polaris CLI
  • PR: #69

0.2.3

🚀 Features

• Add additional metadata to the columns, dataset and benchmark
  • PR: #65

0.2.2

📚 Documentation

• Updated code-example to work out-of-the-box
  • PR: #62

0.2.1

• no changes

0.2.0

🚀 Features

• Refactor the structure of a saved result to be easier for building benchmarks
  • PR: #54
• feat: Update result upload endpoint and payload
  • PR: #56

🐛 Fixes

• fix: Use the Parquet file size and checksum for the table content metadata when creating a dataset on the Hub
  • PR: #57
• Fix issues due to the string representation of the HubOwner
  • PR: #59

0.1.0

🚀 Features

• Redid the BenchmarkResults structure to be tabular
  • PR: #44
• Refactor dataset upload
  • PR: #50
• Extended metrics
  • PR: #48