Nexport

Latest version: v0.4.6

Safety actively analyzes 682229 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 1 of 2

0.4.6

Description

This version of nexport fixes a bug and introduces a new metadata parameter.

Changelog

- Fixed a bug where the experimental json export doesn't comma-separate output neurons.
- Introduces the `activation_function` metadata parameter for the `nexport.export()` function.

0.4.5

Description

This version of nexport fixed an issue with the PyPI distribution requiring the wrong python version.

Changelog

- Fixed PyPI issue.

0.4.3

Description

Woohoo! :tada:

This is the first public release of nexport! Please note this software is still unstable, so there will be bugs.

Changelog

- Added copyright notices & license agreement.

0.4.2

Description

This version of nexport fixes bugs which throw errors when running on supercomputers. Due to the restriction on user access to the OS, export functions which uses OS user credential access results in an error. This version of nexport removes this access requirement and now relies on user input alone.

Changelog

- Fixed an issue with OS user credential access on supercomputers.
- Added metadata object to experimental json export function.

0.4.0

Description

This release of nexport includes an experimental `export_to_json_exprimental()` export function and features a custom JSON encoder which keeps array elements on a single line. This drastically reduces the file length and provides immediate support for [`inference-engine`](https://github.com/BerkeleyLab/inference-engine).

Changelog

`export_to_json_experimental()`

This function uses a custom JSON encoder to achieve array exporting on a single line.

Bugfixes

- Fixed GH-12
- Fixed GH-13

0.3.0

Description

This update brings useful functions for calculating a PyTorch model's number of trainable parameters, layers and neurons. The user has an option to filter the results based on whether they want to include those in the input & output layers.

Changelog

- New functions
- `calculate_layers()`
- `calculate_neurons()`
- `calculate_parameters()`
- Cleaner directory structure
- Example PyTorch networks moved to separate `models.py` file

Page 1 of 2

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.