beanmachine icon indicating copy to clipboard operation
beanmachine copied to clipboard

A library that allows for inference on probabilistic models

Results 81 beanmachine issues
Sort by recently updated
recently updated
newest added

### Motivation Fixes the excess logging for the NMC inferences--both of them--in this tutorial ### Changes proposed Set log level to 50, so only progress bars are shown/logged during NMC...

CLA Signed

Summary: The StatisticalModel class was a class with only static methods. I've removed the class and made the static methods module-level functions. I've also add two "typeguards" to help prove...

CLA Signed
fb-exported

### Motivation Ensure logging is not annoying, add comments to guide users of tutorials, fix a couple typos. ### Changes proposed - remove VerboseLevel.OFF in: Hierarchical_modeling, Hierarchical_regression, Item_Response_Theory, & Zero_inflated_count_data...

CLA Signed

Hi @facebookresearch Love all the work that facebook is doing with probabilistic programming and Bayesian Inference! I'm looking for a speaker to talk about beanmachine and to give an introduction...

### Motivation I hate to do this, but it looks like the compiler upgrade in PyTorch 2.0 somehow broke the pre-existing NNC usage in Bean Machine. Therefore, the algorithms that...

CLA Signed

In file: __init__.py, the list named `__all__` contains an undefined name which can result in errors when this module is imported. I removed the undefined name from the list. Sponsorship...

CLA Signed

In file: __init__.py, the list named `__all__` contains undefined names which can result in errors when this module is imported. I removed the undefined names from the list. For more...

CLA Signed

Summary: I've often said that a big part of writing a compiler is implementing special-purpose hash tables; fortunately this is the first one I've had to make for Beanstalk. I...

CLA Signed
fb-exported

### Issue Description If one uses the function simulate the graphical network is evalatued N times per sample, were N is the number of @bm.random_variable in the network ### Steps...

### Motivation Since there are fewer BM maintainers within Meta, I wanted to open source intern infra projects (by James and Zhen) built on BM for external research and future...

CLA Signed