WIP: This codebase is under active development

Because language models are trained to predict the next token in naturally occurring text, they often reproduce common human errors and misconceptions, even when they "know better" in some sense. More worryingly, when models are trained to generate text that's rated highly by humans, they may learn to output false statements that human evaluators can't detect. We aim to circumvent this issue by directly eliciting latent knowledge (ELK) inside the activations of a language model.

Specifically, we're building on the Contrast Consistent Search (CCS) method described in the paper Discovering Latent Knowledge in Language Models Without Supervision by Burns et al. (2022). In CCS, we search for features in the hidden states of a language model which satisfy certain logical consistency requirements. It turns out that these features are often useful for question-answering and text classification tasks, even though the features are trained without labels.

Our code is based on PyTorch and Huggingface Transformers. We test the code on Python 3.9 and 3.10.

First install the package with pip install -e . in the root directory, or pip install -e .[dev] if you'd like to contribute to the project (see Development section below). This should install all the necessary dependencies.

To extract the hidden states for one model model and the dataset dataset and train a probe on these extracted hidden states, run:

elk elicit microsoft/deberta-v2-xxlarge-mnli imdb --max-examples 1000

To only extract the hidden states for one model model and the dataset dataset, run:

elk extract microsoft/deberta-v2-xxlarge-mnli imdb --max-examples 1000

To only train a CCS model and a logistic regression model

elk train microsoft/deberta-v2-xxlarge-mnli imdb

and evaluate on different datasets: [WIP]

Once finished, results will be saved in ~/.cache/elk/{model}_{prefix}_{seed}.csv

Use pip install pre-commit && pre-commit install in the root folder before your first commit.

If you work on a new feature / fix or some other code task, make sure to create an issue and assign it to yourself (Maybe, even share it in the elk channel of Eleuther's Discord with a small note). In this way, others know you are working on the issue and people won't do the same thing twice 👍 Also others can contact you easily.