v1.0.7

Summary

• Fix TensorFlow STS models failing when pandas supplied float64 by casting every numeric series/covariate to TensorFlow’s active floatx. This keeps HMC/VI and forecasting pipelines in sync and restores TensorFlow runs in notebooks and CI.

• Update project metadata to advertise Python 3.13 support.

Links

Issue: #15 – TensorFlow STS model crashes when covariates/target stay in float64
PR: #16 – Fix TensorFlow dtype mismatch (fixes #15)

v1.0.6

Release v1.0.6: Fix documentation issues and update model defaults

• Fixed Prophet model_args documentation
• Updated all model default values
• Added missing figsize parameter to examples
• Updated summary output format
• Fixes #11 and #12

1.0.1

First release in PiPy