Metadata-Version: 2.1
Name: pytorch-dp
Version: 0.1b1
Summary: Train PyTorch models with Differential Privacy
Home-page: https://github.com/facebookresearch/pytorch-dp
Author: PyTorch Team
License: Apache-2.0
Description: # pytorch-dp: Train PyTorch models with Differential Privacy
        
        [![CircleCI](https://circleci.com/gh/facebookresearch/pytorch-dp.svg?style=svg&circle-token=247aedbb579492ef94652ea93bbee4e169457795)](https://circleci.com/gh/facebookresearch/pytorch-dp)
        
        pytorch-dp is a library that enables training PyTorch models with differential privacy. It supports training with minimal code changes required on the client, has little impact on training performance and allows the client to online track the privacy budget expended at any given moment.
        
        *PyTorch Privacy is currently a preview beta and under active development!*
        
        ### Target audience
        This code release is aimed at two target audiences:
        1. ML practicioners will find this code a gentle introduction to training a model with differential privacy as it requires minimal code changes.
        2. Differential Privacy scientists will find this code easy to experiment and tinker with, allowing them to focus on what matters.
        
        
        ## Installation
        pip:
        ```bash
        pip install pytorch-dp
        ```
        
        From source:
        ```bash
        git clone https://github.com/facebookresearch/pytorch-dp.git
        cd pytorch-dp
        pip install -e .
        ```
        
        ## Getting started
        To train your model with differential privacy, all you need to do is to declare a PrivacyEngine and attach it to your optimizer before running, eg:
        
        ```python
        model = Net()
        optimizer = SGD(model.parameters(), lr=0.05)
        privacy_engine = PrivacyEngine(
            model,
            batch_size,
            sample_size,
            alphas=[1, 10, 100],
            noise_multiplier=1.3,
            max_grad_norm=1.0,
        )
        privacy_engine.attach(optimizer)
        # Now it's business as usual
        ```
        The [MNIST example](examples/mnist.py) contains an end to end run.
        
        ## Contributing
        See the [CONTRIBUTING](CONTRIBUTING.md) file for how to help out.
        
        ## References
        
        * [Mironov, Ilya. "Rényi differential privacy." 2017 IEEE 30th Computer Security Foundations Symposium (CSF). IEEE, 2017.](https://arxiv.org/abs/1702.07476)
        * [Abadi, Martin, et al. "Deep learning with differential privacy." Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security. ACM, 2016.](https://arxiv.org/abs/1607.00133)
        * [Mironov, Ilya, Kunal Talwar, and Li Zhang. "R\'enyi Differential Privacy of the Sampled Gaussian Mechanism." arXiv preprint arXiv:1908.10530 (2019).](https://arxiv.org/abs/1908.10530)
        * [Goodfellow, Ian. "Efficient per-example gradient computations." arXiv preprint arXiv:1510.01799 (2015).](https://arxiv.org/abs/1510.01799)
        * [McMahan, H. Brendan, and Galen Andrew. "A general approach to adding differential privacy to iterative training procedures." arXiv preprint arXiv:1812.06210 (2018).](https://arxiv.org/abs/1812.06210)
        
        ## License
        This code is released under Apache 2.0, as found in the [LICENSE](LICENSE) file.
        
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3
Classifier: Operating System :: OS Independent
Requires-Python: >=3.6
Description-Content-Type: text/markdown
