bayesnf

bayesnf copied to clipboard

fsaad

fsaad

fsaad

Integer division in MAP ensemble_size cause a crash when ensemble_size < device_count

1

https://github.com/google/bayesnf/blob/fb59400ab86aa16a548f6df566bc0d5ba6e19eb5/src/bayesnf/inference.py#L445 If `ensemble_size < jax.device_count` then 0 particles are fitted. In terms of the API `.fit` silently fails, but `.predict` gives an error, since there is a min/max operation over...

fsaad

fsaad

The predicted `mean` array returned by https://github.com/google/bayesnf/blob/2debb69cf21daf7de982843a82436a57d5a97dd3/src/bayesnf/spatiotemporal.py#L273 has three dimensions, (`num_devices, ensemble_size // num_devices, len(table)` (for MLE/MAP) or even four dimensions (for VI). It is more convenient to have only...

fsaad

- For VI, the `ensemble_size` is **per device** - For MAP/MLE, the `ensemble_size` is the total ensemble_size.

fsaad

Bumps the pip group with 1 update in the / directory: [tqdm](https://github.com/tqdm/tqdm). Updates `tqdm` from 4.66.2 to 4.66.3 Release notes Sourced from tqdm's releases. tqdm v4.66.3 stable cli: eval safety...

dependabot[bot]

Follow up questions on making use of seasonality

2

https://google.github.io/bayesnf/tutorials/BayesNF_Tutorial_on_Hungarian_Chickenpox/ ``` from bayesnf.spatiotemporal import BayesianNeuralFieldMAP model = BayesianNeuralFieldMAP( width=256, depth=2, freq='W', seasonality_periods=['M', 'Y'], # equivalent to [365.25/12, 365.25] num_seasonal_harmonics=[2, 10], # two harmonics for M; one harmonic for Y...

f0lie

Bumps the pip group with 2 updates in the / directory: [tqdm](https://github.com/tqdm/tqdm) and [zipp](https://github.com/jaraco/zipp). Updates `tqdm` from 4.66.2 to 4.66.3 Release notes Sourced from tqdm's releases. tqdm v4.66.3 stable cli:...

dependabot[bot]