Do Unlearning Methods Remove Information from Language Model Weights?

Large Language Models' knowledge of how to perform cyber-security attacks, create bioweapons, and manipulate humans poses risks of misuse. Previous work has proposed methods to unlearn this knowledge. Historically, it has been unclear whether unlearning techniques are removing information from the model weights or just making it harder to access. To disentangle these two objectives, we propose an adversarial evaluation method to test for the removal of information from model weights: we give an attacker access to some facts that were supposed to be removed, and using those, the attacker tries to recover other facts from the same distribution that cannot be guessed from the accessible facts. We show that using fine-tuning on the accessible facts can recover 88% of the pre-unlearning accuracy when applied to current unlearning methods, revealing the limitations of these methods in removing information from the model weights.

Stress-Testing Capability Elicitation With Password-Locked Models

To determine the safety of large language models (LLMs), AI developers must be able to assess their dangerous capabilities. But simple prompting strategies often fail to elicit an LLM's full capabilities. One way to elicit capabilities more robustly is to fine-tune the LLM to complete the task. In this paper, we investigate the conditions under which fine-tuning-based elicitation suffices to elicit capabilities. To do this, we introduce password-locked models, LLMs fine-tuned such that some of their capabilities are deliberately hidden. Specifically, these LLMs are trained to exhibit these capabilities only when a password is present in the prompt, and to imitate a much weaker LLM otherwise. Password-locked models enable a novel method of evaluating capabilities elicitation methods, by testing whether these password-locked capabilities can be elicited without using the password. We find that a few high-quality demonstrations are often sufficient to fully elicit password-locked capabilities. More surprisingly, fine-tuning can elicit other capabilities that have been locked using the same password, or even different passwords. Furthermore, when only evaluations, and not demonstrations, are available, approaches like reinforcement learning are still often able to elicit capabilities. Overall, our findings suggest that fine-tuning is an effective method of eliciting hidden capabilities of current models, but may be unreliable when high-quality demonstrations are not available, e.g. as may be the case when models' (hidden) capabilities exceed those of human demonstrators.

AI Control: Improving Safety Despite Intentional Subversion

As large language models (LLMs) become more powerful and are deployed more autonomously, it will be increasingly important to prevent them from causing harmful outcomes. Researchers have investigated a variety of safety techniques for this purpose, e.g. using models to review the outputs of other models, or red-teaming techniques to surface subtle failure modes. However, researchers have not evaluated whether such techniques still ensure safety if the model is itself intentionally trying to subvert them. In this paper, we develop and evaluate pipelines of safety techniques ("protocols") that are robust to intentional subversion. We investigate a scenario in which we want to solve a sequence of programming problems, using access to a powerful but untrusted model (in our case, GPT-4), access to a less powerful trusted model (in our case, GPT-3.5), and limited access to human contractors who provide high-quality trusted labor. We investigate protocols that aim to never submit solutions containing backdoors, which we operationalize here as logical errors that are not caught by test cases. We investigate a range of protocols and test each against strategies that the untrusted model could use to subvert them. One protocol is what we call trusted editing. This protocol first asks GPT-4 to write code, and then asks GPT-3.5 to rate the suspiciousness of that code. If the code is below some suspiciousness threshold, it is submitted. Otherwise, GPT-3.5 edits the solution to remove parts that seem suspicious and then submits the edited code. Another protocol is untrusted monitoring. This protocol asks GPT-4 to write code, and then asks another instance of GPT-4 whether the code is backdoored, using various techniques to prevent the GPT-4 instances from colluding. These protocols improve substantially on simple baselines.

Preventing Language Models From Hiding Their Reasoning

Large language models (LLMs) often benefit from intermediate steps of reasoning to generate answers to complex problems. When these intermediate steps of reasoning are used to monitor the activity of the model, it is essential that this explicit reasoning is faithful, i.e. that it reflects what the model is actually reasoning about. In this work, we focus on one potential way intermediate steps of reasoning could be unfaithful: encoded reasoning, where an LLM could encode intermediate steps of reasoning in the generated text in a way that is not understandable to human readers. We show that language models can be trained to make use of encoded reasoning to get higher performance without the user understanding the intermediate steps of reasoning. We argue that, as language models get stronger, this behavior becomes more likely to appear naturally. Finally, we describe a methodology that enables the evaluation of defenses against encoded reasoning, and show that, under the right conditions, paraphrasing successfully prevents even the best encoding schemes we built from encoding more than 3 bits of information per KB of text.

Benchmarks for Detecting Measurement Tampering

When training powerful AI systems to perform complex tasks, it may be challenging to provide training signals which are robust to optimization. One concern is \textit{measurement tampering}, where the AI system manipulates multiple measurements to create the illusion of good results instead of achieving the desired outcome. In this work, we build four new text-based datasets to evaluate measurement tampering detection techniques on large language models. Concretely, given sets of text inputs and measurements aimed at determining if some outcome occurred, as well as a base model able to accurately predict measurements, the goal is to determine if examples where all measurements indicate the outcome occurred actually had the outcome occur, or if this was caused by measurement tampering. We demonstrate techniques that outperform simple baselines on most datasets, but don't achieve maximum performance. We believe there is significant room for improvement for both techniques and datasets, and we are excited for future work tackling measurement tampering.

Large Language Models Sometimes Generate Purely Negatively-Reinforced Text

When using adversarial training, it is common practice to train against the most egregious failures. However, this might imply using examples with sensitive information (such as leaked passwords or security vulnerabilities) as training data. One might assume that language models trained with gradient descent never generate text snippets which were only present in examples associated with the lowest possible reward. In this paper, we show that this assumption is wrong: in some situations, large language models do learn from such negatively-reinforced examples. We present a specific training setup that enables Pythia-160M to guess passwords 13% more often than it would by guessing randomly, despite only showing it these passwords on examples where the model is incentivized to not output these passwords. Our code is available at www.github.com/FabienRoger/Learning-From-Negative-Examples

Language models are better than humans at next-token prediction

Current language models are considered to have sub-human capabilities at natural language tasks like question-answering or writing code. However, language models are not trained to perform well at these tasks, they are trained to accurately predict the next token given previous tokes in tokenized text. It is not clear whether language models are better or worse than humans at next token prediction. To try to answer this question, we performed two distinct experiments to directly compare humans and language models on this front: one measuring top-1 accuracy and the other measuring perplexity. In both experiments, we find humans to be consistently \emph{worse} than even relatively small language models like GPT3-Ada at next-token prediction.