v0.6.1

Adds a Documentation link (docs site: https://justmytwospence.github.io/cloudposterior/) to the project metadata. No functional changes to the library.

v0.6.0

First PyPI release.

Run PyMC MCMC sampling on cloud VMs with one line:

with cp.cloud(model, remote=True):
    idata = pm.sample()

Automatic caching, live progress, and dashboard/phone notifications included. Supports PyMC 5 (arviz 0.x) and PyMC 6 (arviz 1.x) on Python 3.11-3.13.

pip install cloudposterior

v0.6.0a1

First pre-release: a dry run validating the build and publish pipeline (Trusted Publishing + PEP 740 attestations) before cutting 0.6.0.

v0.4.0

Interactive Dashboard

Live web dashboard for monitoring MCMC sampling remotely.

Features

• Live convergence diagnostics -- R-hat and ESS updated every 50 draws with color-coded status
• Live trace plots -- per-chain traces and posterior KDE with uPlot
• Stop button -- end sampling early with confirmation dialog, keeps draws collected so far
• Light/dark theme -- auto-detects system preference
• QR code -- scan to open dashboard on your phone

Improvements

• Split dashboard and notify into separate parameters (dashboard=True by default for remote)
• Renamed notifications.ipynb to monitoring.ipynb
• Modern card-style layout with responsive breakpoints
• Clean parameter names (model prefix stripped)
• Deferred provisioning -- GPU auto-detected from nuts_sampler at sample time