Pynuml

Latest version: v24.6.0

Safety actively analyzes 638466 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 2 of 2

23.6.0

labelling changes
- minor additions to particle types and physics processes in standard semantic labelling
- added new neutrino flavor event labeller

processing changes
- hit graph processing:
- number of nexus nodes now stored as `num_nodes` attribute instead of empty feature tensor
- changes to truth labels: instead of storing truth labels as `y_f`, `y_s` and `y_i` for filter, semantic and instance labels respectively, we now only store `y_semantic` and `y_instance`. instead of being masked down to only the filtered hits, they are now stored for every hit, and filtered hits are assigned a label of `-1`, which is then filtered during loss calculation using `ignore_index`. the filter labels are now generated by generating a boolean mask, ie. `(y_semantic != -1).float()`.
- updated `labeller` to `semantic_labeller` and added additional `event_labeller` option

other changes
- added `plotly` dependency for graph plotting
- added readthedocs documentation

23.5.2

changes since previous release:
- fix bug in file metadata that slipped thru into a release version
- change "np.int" -> "int" due to recent deprecation of the former in numpy

23.5.1

Patch release to fix PyPi import issues with previous release

23.5.0

This release represents a significant overhaul to essentially all aspects of the package. File IO, truth labelling, object formation and output file standards have all been significantly revised, and graph plotting tools have been added.

0.1.1

Bug fixes:
- File class checks file was successfully opened before trying to close it
- Fixed bug in HDF5 file sequencing and indexing

0.1.0

First PyNuML release, designed for use with MicroBooNE data release

Page 2 of 2

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.