beanmachine
beanmachine copied to clipboard
A library that allows for inference on probabilistic models
### Motivation Continued work on the diagnostics tools. ### Changes proposed - Removed items in the `pointStatistic.ts` module and placed them in other modules for better structure. Updated all modules...
Summary: We currently have `BMGInference`, that takes a BM python program produces a valid BMG graph and runs inference on the BMG C++ backend. In the same vein, this diff...
Differential Revision: D40856513
Summary: The tensorizing / detensorizing code is some of the more complicated code in the compiler and it could benefit from some explanatory comments to demystify it to the new...
Summary: This may lead to more code being type checked than desired, but I think we can use excludes to handle this if that is a problem. Differential Revision: D40606047
Summary: I want to separate out the logic "try to make this node meet a requirement" and "give an error if you cannot"; this will make implementing matrix fill and...
Summary: The requirements-fixing code is not easy to understand, and some small errors and inefficiencies have crept in. I'm going to fix them in this diff to make it then...
Summary: We're continuing to add broadcasting semantics by small steps. In this diff: * I've fixed a bug in how requirements on the matrix add incoming edges are represented. The...
Summary: This diff continues the work to add broadcasting semantics to the compiler and BMG. In this diff, I fix a bug in the type analyzer which was incorrectly detecting...
Summary: Eigen distinguishes between matrices of bools, naturals and doubles; when generating a C++ fragment to construct a BMG graph, we always generated a double matrix, but that's bad; it...