conbo 0.1.1

A package for bounding properties of random variables with high confidence.

ConBo: Bounding properties of random variables with high confidence

ConBo (Confidence Bounds) is the ultimate toolkit for computing distribution-free, non-parametric bounds on properties of random variables with high confidence. It is particularly useful in scenarios where no assumptions can be made about the underlying distribution, for example when working with complex models such as large language models (LLMs) or flow-based generative models.

Installation

To install the required dependencies, run:

pip install conbo

Usage

To use the confidence bounds, import functionality from the conbo module and call the desired functions. Here is an example:

from conbo import *

# Example usage
cdf, cdf_lower, cdf_upper = cdf_bounds(samples, x=[0.5])
sample_mean, exp_lower, exp_upper = expectation_bounds(samples)
sample_variance, var_lower, var_upper = variance_bounds(samples)
sample_std, std_lower, std_upper = std_bounds(samples)

Examples

You can find usage examples in the demo notebook demonstrating how to compute the confidence bounds.

Cite

This repository implements confidence bounds discussed in the paper A Probabilistic Perspective on Unlearning and Alignment for Large Language Models by Yan Scholten, Stephan Günnemann, and Leo Schwinn. Please cite our paper if you use this code in your own work:

@misc{scholten2024probabilistic,
      title={A Probabilistic Perspective on Unlearning and Alignment for Large Language Models},
      author={Yan Scholten and Stephan Günnemann and Leo Schwinn},
      year={2024},
      eprint={2410.03523},
      archivePrefix={arXiv},
      url={https://arxiv.org/abs/2410.03523},
}

Contact

For questions and feedback please contact:

Yan Scholten, Technical University of Munich
Stephan Günnemann, Technical University of Munich
Leo Schwinn, Technical University of Munich

License

This project is licensed under the MIT License. See the LICENSE file for details.