pymc
pymc copied to clipboard
pymc-devs
Bayesian Modeling and Probabilistic Programming in Python
https://github.com/pymc-devs/pymc/blob/dc7cfeeaba98330f2881d627ee988ac267f18df2/pymc/distributions/multivariate.py#L2128 I'm not sure if this is supposed to be a matmul or vecdot. I think the later. If someone is familiar with the distribution please weigh in. CC @bwengals...
## Description This PR updates the PyMC tutorial notebooks with various improvements and fixes to enhance clarity, correctness, and user experience. ## Changes - Fixed outdated code snippets for compatibility...
## Description ## Related Issue - [ ] Closes # - [ ] Related to # ## Checklist - [ ] Checked that [the pre-commit linting/style checks pass](https://docs.pymc.io/en/latest/contributing/python_style.html) - [...
This commit implements validation of InferenceData dimensions and coordinates consistency before pm.sample() executes, preventing cryptic shape mismatch errors during sampling. Changes: - Add pymc/model/validation.py with validation functions: * validate_dims_coords_consistency(): Main...
### Description Make pymc.dims module the default in PyMC 6.0, and move legacy behavior to `pm.numpy`?
This has been around since 2023. Can this be deprecated? https://github.com/pymc-devs/pymc/blob/cd2e1a3979be9a2a27a3d033989db77ebfc97841/pymc/gp/hsgp_approx.py?plain=1#L296-L301
## Description Multiprocess sampling often fails randomly on Linux due to the start method defaulting to "fork". This sets it to "spawn" when `mcmc` is imported. ## Related Issue -...
Improves the clarity and consistency of the dims module notebook in the PyMC docs. Changes * Added cross-references to functions (Deterministic, Potential) and modules * Linked distributions like ZeroSumNormal and...
## Description This PR updates the internal conversion logic to use `dict_to_dataset_drop_incompatible_coords` across all relevant backends. ## Related Issue - [x] Related to #7908 - [x] Closes #7891 ## Checklist...
### Description Samples with multiple steps like Slice could be written to loop in PyTensor. On Numba/JAX backends this should be much faster for models with cheap logp as we...