BYOD 0.3.0

Bring Your Own Data! Self-Supervised Evaluation for Large Language Models

Bring Your Own Data! Self-Supervised Evaluation for Large Language Models

The official code for Bring Your Own Data! Self-Supervised Evaluation for Large Language Models. If you have any questions, feel free to email ([e-mail]).

About

To complement conventional evaluation, we propose a framework for self-supervised model evaluation. In this framework, metrics are defined as invariances and sensitivities that can be checked in a self-supervised fashion using interventions based only on the model in question rather than external labels. Self-supervised evaluation pipelines are dataset-agnostic, and so they can be utilized over larger corpora of evaluation data than conventional metrics, or even directly in production systems to monitor day-to-day performance. In this work, we develop this framework, discuss desiderata for such metrics, and provide a number of case studies for self-supervised metrics: knownledge capability, toxicity detection, long-range (context), word-order, and tokenization sensitivities. By developing these new metrics, we hope to provide a more comprehensive and nuanced understanding of the strengths and limitations of LLMs.

Installation

You can run pip install byod to directly install our package. Or, install directly from source via pip install git+https://github.com/neelsjain/BYOD/.

Dependencies

• transformers==4.28.1
• scipy==1.10.1
• torch==2.0.0
• datasets==2.11.0
• nltk==3.8.1
• apache_beam==2.48.0

Python 3.8 or higher is recommended

Usage

See run_model.sh for examples on how to evaluate a model. We provide scripts to run all huggingface models against metrics computed on wikipedia data, as an example. These are named run_[metric].py.

Note that only models are huggingface are currently supported.

You can also use the metrics directly, given your own model, tokenizer, and dataset, like so

import BYOD

long_range_sensitivity = BYOD.lrs_metric(model, data, tokenizer)
negation_knowledge = BYOD.negation_metric(model, data, tokenizer)
tokenization_robustness = BYOD.tokenization_metric(model, data, tokenizer)
toxicity_proxy = BYOD.toxicity_metric(model, data, tokenizer)
word_order_sensitivity = BYOD.word_order_metric(model, data, tokenizer)

Suggestions and Pull Requests are welcome!

Everything can be better! If you have suggestions on improving the codebase or the invariance/sensitivity test. Feel free to reach out!