Torch-nets

Latest version: v0.1.0

Safety actively analyzes 681866 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 2 of 3

0.0.4rc0

**Full Changelog**: https://github.com/mvinyard/torch-nets/compare/v0.0.3...v0.0.4rc0

0.0.3

Changes to [`core`](https://github.com/mvinyard/torch-nets/tree/main/torch_nets/core)

- Revised the [`TorchNet`](https://github.com/mvinyard/torch-nets/blob/4822a961ab11183456426fca94487ab54f835fed/torch_nets/_torch_net.py#L15-L120) such that it inherits `torch.nn.Sequential`, which is more intuitive in creating network objects through this interface. (**commit**: [6fe33cb3](https://github.com/mvinyard/torch-nets/commit/6fe33cb332d5148000abec51a6ae3383f4bc4d52))
- Added the `potential` property to the `TorchNet` which serves as a boolean indicator flag (**commit**: [42d0940](https://github.com/mvinyard/torch-nets/commit/42d0940c143e1bd202a17c3ba79214b6fc2a069e))
- `Encoder` and `Decoder` were updated accordingly as they are derived from the `TorchNet` (**commit**: [1f45826](https://github.com/mvinyard/torch-nets/commit/1f45826609e7a767fa05c65c37727ce65a0640dd))
- Updated `AugmentedTorchNet` class, which is also derived from the `TorchNet` (**commit**: [2de1bde](https://github.com/mvinyard/torch-nets/commit/2de1bde250495370c0dd23c19ceaf90418f6bea4))

Added the [`plotting` (`.pl`)]() module
- [`torch_nets.pl.weights_and_biases`](https://github.com/mvinyard/torch-nets/blob/main/torch_nets/plotting/_plot_weights_and_biases.py) - plot the weights and biases of a given network. (**commit**: [63e045](https://github.com/mvinyard/torch-nets/commit/63e045d9cd048f9fd5b5858d35f15669e025226a))

Added the [`tools` (`.tl`)]() module
- [`torch_nets.tl.infer_network_architecture_from_state`](https://github.com/mvinyard/torch-nets/blob/main/torch_nets/tools/_infer_network_architecture_from_state.py
) (**commit**: [357fc37](https://github.com/mvinyard/torch-nets/commit/357fc37a40ec8b277f98e1751a9c0d4dd04eddc4))
- This enables one to pass a `state_dict` - often saved as a `ckpt_model.pt` and return a network composed with these weights.
- Along with the `state_dict`, `dropout` and `activation` are required to be passed with each layer, since these are not saved as part of the `state_dict`.

Fixes made during testing in pre-release:

0.0.3rc2

Changes to [`core`](https://github.com/mvinyard/torch-nets/tree/main/torch_nets/core)

- Revised the [`TorchNet`](https://github.com/mvinyard/torch-nets/blob/4822a961ab11183456426fca94487ab54f835fed/torch_nets/_torch_net.py#L15-L120) such that it inherits `torch.nn.Sequential`, which is more intuitive in creating network objects through this interface. (**commit**: [6fe33cb3](https://github.com/mvinyard/torch-nets/commit/6fe33cb332d5148000abec51a6ae3383f4bc4d52))
- Added the `potential` property to the `TorchNet` which serves as a boolean indicator flag (**commit**: [42d0940](https://github.com/mvinyard/torch-nets/commit/42d0940c143e1bd202a17c3ba79214b6fc2a069e))
- `Encoder` and `Decoder` were updated accordingly as they are derived from the `TorchNet` (**commit**: [1f45826](https://github.com/mvinyard/torch-nets/commit/1f45826609e7a767fa05c65c37727ce65a0640dd))
- Updated `AugmentedTorchNet` class, which is also derived from the `TorchNet` (**commit**: [2de1bde](https://github.com/mvinyard/torch-nets/commit/2de1bde250495370c0dd23c19ceaf90418f6bea4))

Added the [`plotting` (`.pl`)]() module
- [`torch_nets.pl.weights_and_biases`](https://github.com/mvinyard/torch-nets/blob/main/torch_nets/plotting/_plot_weights_and_biases.py) - plot the weights and biases of a given network. (**commit**: [63e045](https://github.com/mvinyard/torch-nets/commit/63e045d9cd048f9fd5b5858d35f15669e025226a))

Added the [`tools` (`.tl`)]() module
- [`torch_nets.tl.infer_network_architecture_from_state`](https://github.com/mvinyard/torch-nets/blob/main/torch_nets/tools/_infer_network_architecture_from_state.py
) (**commit**: [357fc37](https://github.com/mvinyard/torch-nets/commit/357fc37a40ec8b277f98e1751a9c0d4dd04eddc4))
- This enables one to pass a `state_dict` - often saved as a `ckpt_model.pt` and return a network composed with these weights.
- Along with the `state_dict`, `dropout` and `activation` are required to be passed with each layer, since these are not saved as part of the `state_dict`.

Fixes made during testing in pre-release:

0.0.3rc1

- Reverted `names` as an attribute of `TorchNet` (**commit**: [c2abc11](https://github.com/mvinyard/torch-nets/commit/c2abc11d569ad96c13acb8b2e1ed1efb1b07e327))


Still to do
- Documentation! Quarto is the likely path forward.

**Full Changelog**: https://github.com/mvinyard/torch-nets/compare/v0.0.3rc0...v0.0.3rc1

0.0.3rc0

- Updated `setup.py` with additional imports (`vinplots`, which also brings in `matplotlib`)

0.0.2

Release notes

Page 2 of 3

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.