Usman Anwar

Stephen Casper, Xander Davies, Claudia Shi et al. · berkeley, eth-zurich

Reinforcement learning from human feedback (RLHF) is a technique for training AI systems to align with human goals. RLHF has emerged as the central method used to finetune state-of-the-art large language models (LLMs). Despite this popularity, there has been relatively little public work systematizing its flaws. In this paper, we (1) survey open problems and fundamental limitations of RLHF and related methods; (2) overview techniques to understand, improve, and complement RLHF in practice; and (3) propose auditing and disclosure standards to improve societal oversight of RLHF systems. Our work emphasizes the limitations of RLHF and highlights the importance of a multi-faceted approach to the development of safer AI systems.

Thomas Coste, Usman Anwar, Robert Kirk et al.

Reinforcement learning from human feedback (RLHF) is a standard approach for fine-tuning large language models to follow instructions. As part of this process, learned reward models are used to approximately model human preferences. However, as imperfect representations of the "true" reward, these learned reward models are susceptible to overoptimization. Gao et al. (2023) studied this phenomenon in a synthetic human feedback setup with a significantly larger "gold" reward model acting as the true reward (instead of humans) and showed that overoptimization remains a persistent problem regardless of the size of the proxy reward model and training data used. Using a similar setup, we conduct a systematic study to evaluate the efficacy of using ensemble-based conservative optimization objectives, specifically worst-case optimization (WCO) and uncertainty-weighted optimization (UWO), for mitigating reward model overoptimization when using two optimization methods: (a) best-of-n sampling (BoN) (b) proximal policy optimization (PPO). We additionally extend the setup of Gao et al. (2023) to include 25% label noise to better mirror real-world conditions. Both with and without label noise, we find that conservative optimization practically eliminates overoptimization and improves performance by up to 70% for BoN sampling. For PPO, ensemble-based conservative optimization always reduces overoptimization and outperforms single reward model optimization. Moreover, combining it with a small KL penalty successfully prevents overoptimization at no performance cost. Overall, our results demonstrate that ensemble-based conservative optimization can effectively counter overoptimization.

Usman Anwar, Julianna Piskorz, David D. Baek et al.

Large language models are beginning to show steganographic capabilities. Such capabilities could allow misaligned models to evade oversight mechanisms. Yet principled methods to detect and quantify such behaviours are lacking. Classical definitions of steganography, and detection methods based on them, require a known reference distribution of non-steganographic signals. For the case of steganographic reasoning in LLMs, knowing such a reference distribution is not feasible; this renders these approaches inapplicable. We propose an alternative, \textbf{decision-theoretic view of steganography}. Our central insight is that steganography creates an asymmetry in usable information between agents who can and cannot decode the hidden content (present within a steganographic signal), and this otherwise latent asymmetry can be inferred from the agents' observable actions. To formalise this perspective, we introduce generalised $\mathcal{V}$-information: a utilitarian framework for measuring the amount of usable information within some input. We use this to define the \textbf{steganographic gap} -- a measure that quantifies steganography by comparing the downstream utility of the steganographic signal to agents that can and cannot decode the hidden content. We empirically validate our formalism, and show that it can be used to detect, quantify, and mitigate steganographic reasoning in LLMs.

Alan Clark, Shoaib Ahmed Siddiqui, Robert Kirk et al.

Existing offline reinforcement learning (RL) algorithms typically assume that training data is either: 1) generated by a known policy, or 2) of entirely unknown origin. We consider multi-demonstrator offline RL, a middle ground where we know which demonstrators generated each dataset, but make no assumptions about the underlying policies of the demonstrators. This is the most natural setting when collecting data from multiple human operators, yet remains unexplored. Since different demonstrators induce different data distributions, we show that this can be naturally framed as a domain generalization problem, with each demonstrator corresponding to a different domain. Specifically, we propose Domain-Invariant Model-based Offline RL (DIMORL), where we apply Risk Extrapolation (REx) (Krueger et al., 2020) to the process of learning dynamics and rewards models. Our results show that models trained with REx exhibit improved domain generalization performance when compared with the natural baseline of pooling all demonstrators' data. We observe that the resulting models frequently enable the learning of superior policies in the offline model-based RL setting, can improve the stability of the policy learning process, and potentially enable increased exploration.

Usman Anwar, Abulhair Saparov, Javier Rando et al. · cambridge, eth-zurich

This work identifies 18 foundational challenges in assuring the alignment and safety of large language models (LLMs). These challenges are organized into three different categories: scientific understanding of LLMs, development and deployment methods, and sociotechnical challenges. Based on the identified challenges, we pose $200+$ concrete research questions.

Usman Anwar, Shehryar Malik, Alireza Aghasi et al.

In real world settings, numerous constraints are present which are hard to specify mathematically. However, for the real world deployment of reinforcement learning (RL), it is critical that RL agents are aware of these constraints, so that they can act safely. In this work, we consider the problem of learning constraints from demonstrations of a constraint-abiding agent's behavior. We experimentally validate our approach and show that our framework can successfully learn the most likely constraints that the agent respects. We further show that these learned constraints are \textit{transferable} to new agents that may have different morphologies and/or reward functions. Previous works in this regard have either mainly been restricted to tabular (discrete) settings, specific types of constraints or assume the environment's transition dynamics. In contrast, our framework is able to learn arbitrary \textit{Markovian} constraints in high-dimensions in a completely model-free setting. The code can be found it: \url{https://github.com/shehryar-malik/icrl}.

Madeline Brumley, Joe Kwon, David Krueger et al.

A key objective of interpretability research on large language models (LLMs) is to develop methods for robustly steering models toward desired behaviors. To this end, two distinct approaches to interpretability -- ``bottom-up" and ``top-down" -- have been presented, but there has been little quantitative comparison between them. We present a case study comparing the effectiveness of representative vector steering methods from each branch: function vectors (FV; arXiv:2310.15213), as a bottom-up method, and in-context vectors (ICV; arXiv:2311.06668) as a top-down method. While both aim to capture compact representations of broad in-context learning tasks, we find they are effective only on specific types of tasks: ICVs outperform FVs in behavioral shifting, whereas FVs excel in tasks requiring more precision. We discuss the implications for future evaluations of steering methods and for further research into top-down and bottom-up steering given these findings.

Thomas Bush, Stephen Chung, Usman Anwar et al.

We present the first mechanistic evidence that model-free reinforcement learning agents can learn to plan. This is achieved by applying a methodology based on concept-based interpretability to a model-free agent in Sokoban -- a commonly used benchmark for studying planning. Specifically, we demonstrate that DRC, a generic model-free agent introduced by Guez et al. (2019), uses learned concept representations to internally formulate plans that both predict the long-term effects of actions on the environment and influence action selection. Our methodology involves: (1) probing for planning-relevant concepts, (2) investigating plan formation within the agent's representations, and (3) verifying that discovered plans (in the agent's representations) have a causal effect on the agent's behavior through interventions. We also show that the emergence of these plans coincides with the emergence of a planning-like property: the ability to benefit from additional test-time compute. Finally, we perform a qualitative analysis of the planning algorithm learned by the agent and discover a strong resemblance to parallelized bidirectional search. Our findings advance understanding of the internal mechanisms underlying planning behavior in agents, which is important given the recent trend of emergent planning and reasoning capabilities in LLMs through RL

Usman Anwar, Johannes Von Oswald, Louis Kirsch et al.

In this work, we make two contributions towards understanding of in-context learning of linear models by transformers. First, we investigate the adversarial robustness of in-context learning in transformers to hijacking attacks -- a type of adversarial attacks in which the adversary's goal is to manipulate the prompt to force the transformer to generate a specific output. We show that both linear transformers and transformers with GPT-2 architectures are vulnerable to such hijacking attacks. However, adversarial robustness to such attacks can be significantly improved through adversarial training -- done either at the pretraining or finetuning stage -- and can generalize to stronger attack models. Our second main contribution is a comparative analysis of adversarial vulnerabilities across transformer models and other algorithms for learning linear models. This reveals two novel findings. First, adversarial attacks transfer poorly between larger transformer models trained from different seeds despite achieving similar in-distribution performance. This suggests that transformers of the same architecture trained according to the same recipe may implement different in-context learning algorithms for the same task. Second, we observe that attacks do not transfer well between classical learning algorithms for linear models (single-step gradient descent and ordinary least squares) and transformers. This suggests that there could be qualitative differences between the in-context learning algorithms that transformers implement and these traditional algorithms.

Aidan Peppin, Anka Reuel, Stephen Casper et al.

To accurately and confidently answer the question 'could an AI model or system increase biorisk', it is necessary to have both a sound theoretical threat model for how AI models or systems could increase biorisk and a robust method for testing that threat model. This paper provides an analysis of existing available research surrounding two AI and biorisk threat models: 1) access to information and planning via large language models (LLMs), and 2) the use of AI-enabled biological tools (BTs) in synthesizing novel biological artifacts. We find that existing studies around AI-related biorisk are nascent, often speculative in nature, or limited in terms of their methodological maturity and transparency. The available literature suggests that current LLMs and BTs do not pose an immediate risk, and more work is needed to develop rigorous approaches to understanding how future models could increase biorisks. We end with recommendations about how empirical work can be expanded to more precisely target biorisk and ensure rigor and validity of findings.

Jose Miguel Lara Rangel, Stefan Schoepf, Jack Foster et al.

Machine unlearning is gaining increasing attention as a way to remove adversarial data poisoning attacks from already trained models and to comply with privacy and AI regulations. The objective is to unlearn the effect of undesired data from a trained model while maintaining performance on the remaining data. This paper introduces HyperForget, a novel machine unlearning framework that leverages hypernetworks - neural networks that generate parameters for other networks - to dynamically sample models that lack knowledge of targeted data while preserving essential capabilities. Leveraging diffusion models, we implement two Diffusion HyperForget Networks and used them to sample unlearned models in Proof-of-Concept experiments. The unlearned models obtained zero accuracy on the forget set, while preserving good accuracy on the retain sets, highlighting the potential of HyperForget for dynamic targeted data removal and a promising direction for developing adaptive machine unlearning algorithms.

Usman Anwar, Tim Bakker, Dana Kianfar et al.

Chain-of-thought (CoT) monitors are LLM-based systems that analyze reasoning traces to detect when outputs may exhibit attributes of interest, such as test-hacking behavior during code generation. In this paper, we use information-theoretic analysis to show that non-zero mutual information between CoT and output is a necessary but not sufficient condition for CoT monitorability. We identify two sources of approximation error that may undermine the performance of CoT monitors in practice: information gap, which measures the extent to which the monitor can extract the information available in CoT, and elicitation error, which measures the extent to which the monitor approximates the optimal monitoring function. We further demonstrate that CoT monitorability can be systematically improved through targeted training objectives. To this end, we propose two complementary approaches: (a) an oracle-based method that directly rewards the monitored model for producing CoTs that maximize monitor accuracy, and (b) a more practical, label-free approach that maximizes conditional mutual information between outputs and CoTs. Across multiple different environments, we show both methods significantly improve monitor accuracy while preventing CoT degeneration even when training against a monitor, thereby mitigating reward hacking when the task reward is imperfectly specified.

Karim Abdel Sadek, Matthew Farrugia-Roberts, Usman Anwar et al. · cambridge, gatech

Safe generalization in reinforcement learning requires not only that a learned policy acts capably in new situations, but also that it uses its capabilities towards the pursuit of the designer's intended goal. The latter requirement may fail when a proxy goal incentivizes similar behavior to the intended goal within the training environment, but not in novel deployment environments. This creates the risk that policies will behave as if in pursuit of the proxy goal, rather than the intended goal, in deployment -- a phenomenon known as goal misgeneralization. In this paper, we formalize this problem setting in order to theoretically study the possibility of goal misgeneralization under different training objectives. We show that goal misgeneralization is possible under approximate optimization of the maximum expected value (MEV) objective, but not the minimax expected regret (MMER) objective. We then empirically show that the standard MEV-based training method of domain randomization exhibits goal misgeneralization in procedurally-generated grid-world environments, whereas current regret-based unsupervised environment design (UED) methods are more robust to goal misgeneralization (though they don't find MMER policies in all cases). Our findings suggest that minimax expected regret is a promising approach to mitigating goal misgeneralization.

Usman Anwar, Ashish Pandian, Jia Wan et al.

Zero-shot coordination (ZSC) is a popular setting for studying the ability of reinforcement learning (RL) agents to coordinate with novel partners. Prior ZSC formulations assume the $\textit{problem setting}$ is common knowledge: each agent knows the underlying Dec-POMDP, knows others have this knowledge, and so on ad infinitum. However, this assumption rarely holds in complex real-world settings, which are often difficult to fully and correctly specify. Hence, in settings where this common knowledge assumption is invalid, agents trained using ZSC methods may not be able to coordinate well. To address this limitation, we formulate the $\textit{noisy zero-shot coordination}$ (NZSC) problem. In NZSC, agents observe different noisy versions of the ground truth Dec-POMDP, which are assumed to be distributed according to a fixed noise model. Only the distribution of ground truth Dec-POMDPs and the noise model are common knowledge. We show that a NZSC problem can be reduced to a ZSC problem by designing a meta-Dec-POMDP with an augmented state space consisting of all the ground-truth Dec-POMDPs. For solving NZSC problems, we propose a simple and flexible meta-learning method called NZSC training, in which the agents are trained across a distribution of coordination problems - which they only get to observe noisy versions of. We show that with NZSC training, RL agents can be trained to coordinate well with novel partners even when the (exact) problem setting of the coordination is not common knowledge.

Alan Chan, Noam Kolt, Peter Wills et al.

AI systems are increasingly pervasive, yet information needed to decide whether and how to engage with them may not exist or be accessible. A user may not be able to verify whether a system has certain safety certifications. An investigator may not know whom to investigate when a system causes an incident. It may not be clear whom to contact to shut down a malfunctioning system. Across a number of domains, IDs address analogous problems by identifying particular entities (e.g., a particular Boeing 747) and providing information about other entities of the same class (e.g., some or all Boeing 747s). We propose a framework in which IDs are ascribed to instances of AI systems (e.g., a particular chat session with Claude 3), and associated information is accessible to parties seeking to interact with that system. We characterize IDs for AI systems, provide concrete examples where IDs could be useful, argue that there could be significant demand for IDs from key actors, analyze how those actors could incentivize ID adoption, explore a potential implementation of our framework for deployers of AI systems, and highlight limitations and risks. IDs seem most warranted in settings where AI systems could have a large impact upon the world, such as in making financial transactions or contacting real humans. With further study, IDs could help to manage a world where AI systems pervade society.

Usman Anwar, Tughrul Arslan, Amir Hussain

This paper provides an overview of hearing loss effects on neurological function and progressive diseases; and explores the role of cognitive load monitoring to detect dementia. It also investigates the prospects of utilizing hearing aid technology to reverse cognitive decline and delay the onset of dementia, for the old age population. The interrelation between hearing loss, cognitive load and dementia is discussed. Future considerations for improvement with respect to robust diagnosis, user centricity, device accuracy and privacy for wider clinical practice is also explored. The review concludes by discussing the future scope and potential of designing practical wearable microwave technologies and evaluating their use in smart care homes setting.

Shehryar Malik, Usman Anwar, Ali Ahmed et al.

Recently, there has been a lot of interest in using neural networks for solving partial differential equations. A number of neural network-based partial differential equation solvers have been formulated which provide performances equivalent, and in some cases even superior, to classical solvers. However, these neural solvers, in general, need to be retrained each time the initial conditions or the domain of the partial differential equation changes. In this work, we posit the problem of approximating the solution of a fixed partial differential equation for any arbitrary initial conditions as learning a conditional probability distribution. We demonstrate the utility of our method on Burger's Equation.