Tristan Hume

Nelson Elhage, Tristan Hume, Catherine Olsson et al. · openai

Neural networks often pack many unrelated concepts into a single neuron - a puzzling phenomenon known as 'polysemanticity' which makes interpretability much more challenging. This paper provides a toy model where polysemanticity can be fully understood, arising as a result of models storing additional sparse features in "superposition." We demonstrate the existence of a phase change, a surprising connection to the geometry of uniform polytopes, and evidence of a link to adversarial examples. We also discuss potential implications for mechanistic interpretability.

Adly Templeton, Tom Conerly, Jonathan Marcus et al.

We demonstrate that sparse autoencoders can extract interpretable features from Claude 3 Sonnet, a production-scale language model, addressing the open question of whether dictionary learning methods scale beyond small transformers. We trained sparse autoencoders with up to 34 million features on the model's middle layer residual stream, using scaling laws to guide hyperparameter selection. The resulting features are multilingual and multimodal (generalizing to images despite text-only training), respond to both concrete instances and abstract discussions of concepts, and can be used to steer model behavior in ways consistent with their interpretations. We find features corresponding to famous entities and locations, as well as more abstract concepts like sarcasm or errors in code. We also identify features relevant to ways in which language models might cause harm--including features representing deception, power-seeking, sycophancy, and bias--and show that these causally influence model outputs when manipulated. Additionally, we conduct analyses of feature interpretability, geometry, and computational function. However, significant limitations remain: our suite of features is incomplete, and we lack rigorous methods for evaluating whether our features faithfully capture model computations.

Yuntao Bai, Saurav Kadavath, Sandipan Kundu et al. · anthropic, berkeley

As AI systems become more capable, we would like to enlist their help to supervise other AIs. We experiment with methods for training a harmless AI assistant through self-improvement, without any human labels identifying harmful outputs. The only human oversight is provided through a list of rules or principles, and so we refer to the method as 'Constitutional AI'. The process involves both a supervised learning and a reinforcement learning phase. In the supervised phase we sample from an initial model, then generate self-critiques and revisions, and then finetune the original model on revised responses. In the RL phase, we sample from the finetuned model, use a model to evaluate which of the two samples is better, and then train a preference model from this dataset of AI preferences. We then train with RL using the preference model as the reward signal, i.e. we use 'RL from AI Feedback' (RLAIF). As a result we are able to train a harmless but non-evasive AI assistant that engages with harmful queries by explaining its objections to them. Both the SL and RL methods can leverage chain-of-thought style reasoning to improve the human-judged performance and transparency of AI decision making. These methods make it possible to control AI behavior more precisely and with far fewer human labels.

Yuntao Bai, Andy Jones, Kamal Ndousse et al. · berkeley, openai

We apply preference modeling and reinforcement learning from human feedback (RLHF) to finetune language models to act as helpful and harmless assistants. We find this alignment training improves performance on almost all NLP evaluations, and is fully compatible with training for specialized skills such as python coding and summarization. We explore an iterated online mode of training, where preference models and RL policies are updated on a weekly cadence with fresh human feedback data, efficiently improving our datasets and models. Finally, we investigate the robustness of RLHF training, and identify a roughly linear relation between the RL reward and the square root of the KL divergence between the policy and its initialization. Alongside our main results, we perform peripheral analyses on calibration, competing objectives, and the use of OOD detection, compare our models with human writers, and provide samples from our models using prompts appearing in recent related work.

Saurav Kadavath, Tom Conerly, Amanda Askell et al. · berkeley, openai

We study whether language models can evaluate the validity of their own claims and predict which questions they will be able to answer correctly. We first show that larger models are well-calibrated on diverse multiple choice and true/false questions when they are provided in the right format. Thus we can approach self-evaluation on open-ended sampling tasks by asking models to first propose answers, and then to evaluate the probability "P(True)" that their answers are correct. We find encouraging performance, calibration, and scaling for P(True) on a diverse array of tasks. Performance at self-evaluation further improves when we allow models to consider many of their own samples before predicting the validity of one specific possibility. Next, we investigate whether models can be trained to predict "P(IK)", the probability that "I know" the answer to a question, without reference to any particular proposed answer. Models perform well at predicting P(IK) and partially generalize across tasks, though they struggle with calibration of P(IK) on new tasks. The predicted P(IK) probabilities also increase appropriately in the presence of relevant source materials in the context, and in the presence of hints towards the solution of mathematical word problems. We hope these observations lay the groundwork for training more honest models, and for investigating how honesty generalizes to cases where models are trained on objectives other than the imitation of human writing.

Deep Ganguli, Liane Lovitt, Jackson Kernion et al. · berkeley, openai

We describe our early efforts to red team language models in order to simultaneously discover, measure, and attempt to reduce their potentially harmful outputs. We make three main contributions. First, we investigate scaling behaviors for red teaming across 3 model sizes (2.7B, 13B, and 52B parameters) and 4 model types: a plain language model (LM); an LM prompted to be helpful, honest, and harmless; an LM with rejection sampling; and a model trained to be helpful and harmless using reinforcement learning from human feedback (RLHF). We find that the RLHF models are increasingly difficult to red team as they scale, and we find a flat trend with scale for the other model types. Second, we release our dataset of 38,961 red team attacks for others to analyze and learn from. We provide our own analysis of the data and find a variety of harmful outputs, which range from offensive language to more subtly harmful non-violent unethical outputs. Third, we exhaustively describe our instructions, processes, statistical methodologies, and uncertainty about red teaming. We hope that this transparency accelerates our ability to work together as a community in order to develop shared norms, practices, and technical standards for how to red team language models.

Ethan Perez, Sam Ringer, Kamilė Lukošiūtė et al. · anthropic, berkeley

As language models (LMs) scale, they develop many novel behaviors, good and bad, exacerbating the need to evaluate how they behave. Prior work creates evaluations with crowdwork (which is time-consuming and expensive) or existing data sources (which are not always available). Here, we automatically generate evaluations with LMs. We explore approaches with varying amounts of human effort, from instructing LMs to write yes/no questions to making complex Winogender schemas with multiple stages of LM-based generation and filtering. Crowdworkers rate the examples as highly relevant and agree with 90-100% of labels, sometimes more so than corresponding human-written datasets. We generate 154 datasets and discover new cases of inverse scaling where LMs get worse with size. Larger LMs repeat back a dialog user's preferred answer ("sycophancy") and express greater desire to pursue concerning goals like resource acquisition and goal preservation. We also find some of the first examples of inverse scaling in RL from Human Feedback (RLHF), where more RLHF makes LMs worse. For example, RLHF makes LMs express stronger political views (on gun rights and immigration) and a greater desire to avoid shut down. Overall, LM-written evaluations are high-quality and let us quickly discover many novel LM behaviors.

Tamera Lanham, Anna Chen, Ansh Radhakrishnan et al. · berkeley, stanford

Large language models (LLMs) perform better when they produce step-by-step, "Chain-of-Thought" (CoT) reasoning before answering a question, but it is unclear if the stated reasoning is a faithful explanation of the model's actual reasoning (i.e., its process for answering the question). We investigate hypotheses for how CoT reasoning may be unfaithful, by examining how the model predictions change when we intervene on the CoT (e.g., by adding mistakes or paraphrasing it). Models show large variation across tasks in how strongly they condition on the CoT when predicting their answer, sometimes relying heavily on the CoT and other times primarily ignoring it. CoT's performance boost does not seem to come from CoT's added test-time compute alone or from information encoded via the particular phrasing of the CoT. As models become larger and more capable, they produce less faithful reasoning on most tasks we study. Overall, our results suggest that CoT can be faithful if the circumstances such as the model size and task are carefully chosen.

Deep Ganguli, Amanda Askell, Nicholas Schiefer et al. · berkeley, openai

We test the hypothesis that language models trained with reinforcement learning from human feedback (RLHF) have the capability to "morally self-correct" -- to avoid producing harmful outputs -- if instructed to do so. We find strong evidence in support of this hypothesis across three different experiments, each of which reveal different facets of moral self-correction. We find that the capability for moral self-correction emerges at 22B model parameters, and typically improves with increasing model size and RLHF training. We believe that at this level of scale, language models obtain two capabilities that they can use for moral self-correction: (1) they can follow instructions and (2) they can learn complex normative concepts of harm like stereotyping, bias, and discrimination. As such, they can follow instructions to avoid certain kinds of morally harmful outputs. We believe our results are cause for cautious optimism regarding the ability to train language models to abide by ethical principles.

Samuel R. Bowman, Jeeyoon Hyun, Ethan Perez et al. · anthropic, openai

Developing safe and useful general-purpose AI systems will require us to make progress on scalable oversight: the problem of supervising systems that potentially outperform us on most skills relevant to the task at hand. Empirical work on this problem is not straightforward, since we do not yet have systems that broadly exceed our abilities. This paper discusses one of the major ways we think about this problem, with a focus on ways it can be studied empirically. We first present an experimental design centered on tasks for which human specialists succeed but unaided humans and current general AI systems fail. We then present a proof-of-concept experiment meant to demonstrate a key feature of this experimental design and show its viability with two question-answering tasks: MMLU and time-limited QuALITY. On these tasks, we find that human participants who interact with an unreliable large-language-model dialog assistant through chat -- a trivial baseline strategy for scalable oversight -- substantially outperform both the model alone and their own unaided performance. These results are an encouraging sign that scalable oversight will be tractable to study with present models and bolster recent findings that large language models can productively assist humans with difficult tasks.

Danny Hernandez, Tom Brown, Tom Conerly et al. · openai

Recent large language models have been trained on vast datasets, but also often on repeated data, either intentionally for the purpose of upweighting higher quality data, or unintentionally because data deduplication is not perfect and the model is exposed to repeated data at the sentence, paragraph, or document level. Some works have reported substantial negative performance effects of this repeated data. In this paper we attempt to study repeated data systematically and to understand its effects mechanistically. To do this, we train a family of models where most of the data is unique but a small fraction of it is repeated many times. We find a strong double descent phenomenon, in which repeated data can lead test loss to increase midway through training. A predictable range of repetition frequency leads to surprisingly severe degradation in performance. For instance, performance of an 800M parameter model can be degraded to that of a 2x smaller model (400M params) by repeating 0.1% of the data 100 times, despite the other 90% of the training tokens remaining unique. We suspect there is a range in the middle where the data can be memorized and doing so consumes a large fraction of the model's capacity, and this may be where the peak of degradation occurs. Finally, we connect these observations to recent mechanistic interpretability work - attempting to reverse engineer the detailed computations performed by the model - by showing that data repetition disproportionately damages copying and internal structures associated with generalization, such as induction heads, providing a possible mechanism for the shift from generalization to memorization. Taken together, these results provide a hypothesis for why repeating a relatively small fraction of data in large language models could lead to disproportionately large harms to performance.

Sandipan Kundu, Yuntao Bai, Saurav Kadavath et al. · berkeley, openai

Human feedback can prevent overtly harmful utterances in conversational models, but may not automatically mitigate subtle problematic behaviors such as a stated desire for self-preservation or power. Constitutional AI offers an alternative, replacing human feedback with feedback from AI models conditioned only on a list of written principles. We find this approach effectively prevents the expression of such behaviors. The success of simple principles motivates us to ask: can models learn general ethical behaviors from only a single written principle? To test this, we run experiments using a principle roughly stated as "do what's best for humanity". We find that the largest dialogue models can generalize from this short constitution, resulting in harmless assistants with no stated interest in specific motivations like power. A general principle may thus partially avoid the need for a long list of constitutions targeting potentially harmful behaviors. However, more detailed constitutions still improve fine-grained control over specific types of harms. This suggests both general and specific principles have value for steering AI safely.