While dense retrieval has been shown effective and efficient across tasks and languages, it remains difficult to create effective fully zero-shot dense retrieval systems when no relevance label is available. In this paper, we recognize the difficulty of zero-shot learning and encoding relevance. Instead, we propose to pivot through Hypothetical Document Embeddings~(HyDE). Given a query, HyDE first zero-shot instructs an instruction-following language model (e.g. InstructGPT) to generate a hypothetical document. The document captures relevance patterns but is unreal and may contain false details. Then, an unsupervised contrastively learned encoder~(e.g. Contriever) encodes the document into an embedding vector. This vector identifies a neighborhood in the corpus embedding space, where similar real documents are retrieved based on vector similarity. This second step ground the generated document to the actual corpus, with the encoder's dense bottleneck filtering out the incorrect details. Our experiments show that HyDE significantly outperforms the state-of-the-art unsupervised dense retriever Contriever and shows strong performance comparable to fine-tuned retrievers, across various tasks (e.g. web search, QA, fact verification) and languages~(e.g. sw, ko, ja).

Systems for knowledge-intensive tasks such as open-domain question answering (QA) usually consist of two stages: efficient retrieval of relevant documents from a large corpus and detailed reading of the selected documents to generate answers. Retrievers and readers are usually modeled separately, which necessitates a cumbersome implementation and is hard to train and adapt in an end-to-end fashion. In this paper, we revisit this design and eschew the separate architecture and training in favor of a single Transformer that performs Retrieval as Attention (ReAtt), and end-to-end training solely based on supervision from the end QA task. We demonstrate for the first time that a single model trained end-to-end can achieve both competitive retrieval and QA performance, matching or slightly outperforming state-of-the-art separately trained retrievers and readers. Moreover, end-to-end adaptation significantly boosts its performance on out-of-domain datasets in both supervised and unsupervised settings, making our model a simple and adaptable solution for knowledge-intensive tasks. Code and models are available at https://github.com/jzbjyb/ReAtt.

ClueWeb22, the newest iteration of the ClueWeb line of datasets, provides 10 billion web pages affiliated with rich information. Its design was influenced by the need for a high quality, large scale web corpus to support a range of academic and industry research, for example, in information systems, retrieval-augmented AI systems, and model pretraining. Compared with earlier ClueWeb corpora, the ClueWeb22 corpus is larger, more varied, of higher-quality, and aligned with the document distributions in commercial web search. Besides raw HTML, ClueWeb22 includes rich information about the web pages provided by industry-standard document understanding systems, including the visual representation of pages rendered by a web browser, parsed HTML structure information from a neural network parser, and pre-processed cleaned document text to lower the barrier to entry. Many of these signals have been widely used in industry but are available to the research community for the first time at this scale.

Wearable sensors for measuring head kinematics can be noisy due to imperfect interfaces with the body. Mouthguards are used to measure head kinematics during impacts in traumatic brain injury (TBI) studies, but deviations from reference kinematics can still occur due to potential looseness. In this study, deep learning is used to compensate for the imperfect interface and improve measurement accuracy. A set of one-dimensional convolutional neural network (1D-CNN) models was developed to denoise mouthguard kinematics measurements along three spatial axes of linear acceleration and angular velocity. The denoised kinematics had significantly reduced errors compared to reference kinematics, and reduced errors in brain injury criteria and tissue strain and strain rate calculated via finite element modeling. The 1D-CNN models were also tested on an on-field dataset of college football impacts and a post-mortem human subject dataset, with similar denoising effects observed. The models can be used to improve detection of head impacts and TBI risk evaluation, and potentially extended to other sensors measuring kinematics.

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.

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.

We present a method for providing statistical guarantees on runtime safety and goal reachability for integrated planning and control of a class of systems with unknown nonlinear stochastic underactuated dynamics. Specifically, given a dynamics dataset, our method jointly learns a mean dynamics model, a spatially-varying disturbance bound that captures the effect of noise and model mismatch, and a feedback controller based on contraction theory that stabilizes the learned dynamics. We propose a sampling-based planner that uses the mean dynamics model and simultaneously bounds the closed-loop tracking error via a learned disturbance bound. We employ techniques from Extreme Value Theory (EVT) to estimate, to a specified level of confidence, several constants which characterize the learned components and govern the size of the tracking error bound. This ensures plans are guaranteed to be safely tracked at runtime. We validate that our guarantees translate to empirical safety in simulation on a 10D quadrotor, and in the real world on a physical CrazyFlie quadrotor and Clearpath Jackal robot, whereas baselines that ignore the model error and stochasticity are unsafe.

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 how to turn it into one that can be productively studied empirically. We first present an experimental design centered on choosing tasks for which human specialists succeed but unaided humans and current general AI systems fail. We then present a proof-of-concept experiment following 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.

在机器学习中，对神经网络集合（NNE）（NNE）引起了新的兴趣，从而从一组较小的模型（而不是从单个较大的模型）中获得了预测作为汇总的预测。在这里，我们展示了如何使用随机系统中稀有轨迹的技术来定义和训练NNE。我们根据模型参数的轨迹定义一个NNE，在简单的，离散的时间，扩散动力学下，并通过将这些轨迹偏向较小的时间整合损失来训练NNE，并由适当的计数领域控制，这些领域的作用是超参数。我们证明了该技术在一系列简单监督的学习任务上的生存能力。与更常规的基于梯度的方法相比，我们讨论了轨迹采样方法的潜在优势。

我们展示了如何采用回归函数$ \ hat {f} $，该{f} $适当地``多校准''并有效地将其后处理成近似错误的分类器，使分类器满足各种公平限制。后处理不需要标记的数据，只有一定数量的未标记数据和计算。计算$ \ hat f $的计算和样本复杂性要求与解决单个公平学习任务的要求相媲美，但实际上可以用来有效地解决许多不同的下游公平约束的学习问题。我们的后处理方法可以轻松处理相交组，从而将先前的工作推广到后处理回归功能上，以满足仅应用于分离组的公平约束。我们的工作扩展了最近的工作，表明多校准的回归函数是``omnipredictors''（即可以在后处理以最佳解决无约束的ERM问题）以进行约束优化。