在改善的核心，会话AI是如何评估对话的公开问题。具有自动指标的问题是众所周知的（Liu等，2016年，Arxiv：1603.08023），人类评估仍然认为黄金标准。不幸的是，如何进行人类评估也是一个公开问题：不同的数据收集方法具有不同程度的人类协议和统计敏感性，导致人类注释时间和劳动力成本不同。在这项工作中，我们比较五个不同的人群人的人类评估方法，并发现不同的方法是最重要的，具体取决于模型的类型相比，董事会没有明确的赢家。虽然这突出了该地区的开放问题，但我们的分析导致建议何时使用哪一个以及未来的未来方向。

There has been great recent advancement in human-computer chat. However, proper evaluation currently requires human judgements that produce notoriously high-variance metrics due to their inherent subjectivity. Furthermore, there is little standardization in the methods and labels used for evaluation, with an overall lack of work to compare and assess the validity of various evaluation approaches. As a consequence, existing evaluation results likely leave an incomplete picture of the strengths and weaknesses of open-domain chatbots. We aim towards a dimensional evaluation of human-computer chat that can reliably measure several distinct aspects of chat quality. To this end, we present our novel human evaluation method that quantifies the rate of several quality-related chatbot behaviors. Our results demonstrate our method to be more suitable for dimensional chat evaluation than alternative likert-style or comparative methods. We then use our validated method and existing methods to evaluate four open-domain chat models from the recent literature.

我们介绍了Sparrow，这是一个寻求信息的对话代理，与提示的语言模型基线相比，训练有素，更有帮助，正确和无害。我们使用从人类反馈中的强化学习来培训我们的模型，以帮助人类评估者判断代理人的行为。首先，为了使我们的代理人更有帮助和无害，我们将良好对话的要求分解为代理人应遵循的自然语言规则，并分别向评估者询问每个规则。我们证明，这种崩溃使我们能够收集对代理行为的更多针对性的人类判断，并允许更有效的规则条件奖励模型。其次，我们的代理商在收集对模型声明的偏好判决时提供了支持事实主张的来源的证据。对于事实问题，麻雀提供的证据支持了78％的时间。比基线比基线更享受麻雀，同时对人类的对抗性探测更具弹性，在探测时只有8％的时间违反了我们的规则。最后，我们进行了广泛的分析，表明尽管我们的模型学会遵守我们的规则，但它可以表现出分布偏见。

Chit-chat models are known to have several problems: they lack specificity, do not display a consistent personality and are often not very captivating. In this work we present the task of making chit-chat more engaging by conditioning on profile information. We collect data and train models to (i) condition on their given profile information; and (ii) information about the person they are talking to, resulting in improved dialogues, as measured by next utterance prediction. Since (ii) is initially unknown, our model is trained to engage its partner with personal topics, and we show the resulting dialogue can be used to predict profile information about the interlocutors.

In open-domain dialogue intelligent agents should exhibit the use of knowledge, however there are few convincing demonstrations of this to date. The most popular sequence to sequence models typically "generate and hope" generic utterances that can be memorized in the weights of the model when mapping from input utterance(s) to output, rather than employing recalled knowledge as context. Use of knowledge has so far proved difficult, in part because of the lack of a supervised learning benchmark task which exhibits knowledgeable open dialogue with clear grounding. To that end we collect and release a large dataset with conversations directly grounded with knowledge retrieved from Wikipedia. We then design architectures capable of retrieving knowledge, reading and conditioning on it, and finally generating natural responses. Our best performing dialogue models are able to conduct knowledgeable discussions on open-domain topics as evaluated by automatic metrics and human evaluations, while our new benchmark allows for measuring further improvements in this important research direction.

我们提出了Blenderbot 3，这是一个175B参数对话模型，能够通过访问Internet和长期内存进行开放域对话，并接受了大量用户定义的任务的培训。我们同时发布了模型权重和代码，还将模型部署在公共网页上，以与有机用户进行交互。该技术报告描述了该模型的构建方式（建筑，模型和培训计划）以及其部署的细节，包括安全机制。人类评估表明，它优于现有的开放域对话代理，包括其前身（Roller等，2021； Komeili等，2022）。最后，我们使用部署收集的数据详细介绍了持续学习的计划，该数据也将公开发布。因此，该研究计划的目标是使社区能够研究通过互动学习的不断改进的负责任的代理商。

随着近期自然语言生成（NLG）模型的各种应用程序的改进，它变得必须具有识别和评估NLG输出是否仅共享关于外部世界的可验证信息的手段。在这项工作中，我们提出了一个归属于识别的来源（AIS）的新评估框架，用于评估自然语言生成模型的输出，当这种输出涉及外部世界时。我们首先定义AIS，并引入两级注释管道，用于允许注释器根据AIS指南适当地评估模型输出。通过人为评估研究，我们在三个代数据集（会话QA域中的两个中和总结一下，概括地验证了这种方法，表明AIS可以作为测量模型生成的语句是否支持基础来源的常见框架。我们释放人类评估研究指南。

经过培训的模拟静态数据集的冷冻模型永远无法提高其性能。可以采用互联网进行互联网以获取最新信息并在部署期间从人类那里获得反馈的模型提供了适应新信息并提高其性能的承诺。在这项工作中，我们研究了如何在此类学习框架中提高以互联网为导向的对话技能。我们收集人类互动的部署数据，并公开可用，并收集各种类型的人类反馈 - 包括二进制质量测量，自由形式的文本反馈和罚款良好的失败原因。然后，我们研究了各种从此类反馈中改进的算法，包括标准监督学习，拒绝抽样，模型引导和基于奖励的学习，以便对哪种类型的反馈和算法效果最好。我们发现最近介绍的导演模型（Arora等人，'22）比其他现有方法显示出显着改善。

我们介绍了Godel（接地开放对话语言模型），这是对话框的大型预训练的语言模型。与诸如Dialogpt之类的早期模型相比，Godel利用了一个新的扎根预训练阶段，旨在更好地支持将Godel适应广泛的下游对话框任务，这些任务需要当前对话外部的信息（例如，数据库或文档）到产生良好的回应。针对一系列基准测试的实验，这些基准涵盖了面向任务的对话框，对话质量质量检查和接地的开放式对话框，表明Godel在几次以上的微调设置中优于最先进的预训练的对话模型，就人类和自动评估。我们评估方法的一个新颖特征是引入了一个效用概念，该概念除了其交流特征（内在评估）外，还评估了响应的有用性（外部评估）。我们表明，外部评估提供了改进的通道间一致性和与自动指标的相关性。代码和数据处理脚本公开可用。

对话研究的最终目标是开发可以在交互式设置中有效使用的系统。为此，我们在第9对话系统技术挑战中介绍了对话框的交互式评估。该曲目由两个子任务组成。第一个子任务涉及建立知识接地的响应生成模型。第二个子任务旨在通过与真实用户的交互式设置进行评估，旨在将对话模型扩展到静态数据集之外。我们的曲目挑战参与者开发强大的响应生成模型，并探索将它们扩展到与真实用户的来回互动的策略。从静态语料库到交互式评估的发展引入了独特的挑战，并促进了对开放域对话系统的更全面评估。本文概述了曲目，包括方法和结果。此外，它提供了有关如何最佳评估开放域对话框模型的见解

鉴于大型语言模型的广泛能力，应该有可能朝着一般的文本的助手工作，这些助手与人类价值一致，这意味着它是有帮助，诚实的和无害的。在此方向上的初始遗传，我们研究简单的基线技术和评估，例如提示。我们发现，从模型规模增加适度的干预措施的好处，概括为各种对准评估，并不会损害大型模型的性能。接下来，我们调查与对齐，比较仿制，二进制歧视和排名偏好建模相关的几个培训目标的缩放趋势。我们发现排名优先级模型比模仿学习更好地表现得多，并且通常以模型大小更有利地缩放。相比之下，二进制歧视通常与模仿学习非常类似地执行和缩放。最后，我们研究了一种“偏好模型预训练阶段的培训阶段，其目的是在对人偏好的芬明时提高样本效率。

Many real-world applications of language models (LMs), such as code autocomplete and writing assistance, involve human-LM interaction, but the main LM benchmarks are non-interactive, where a system produces output without human intervention. To evaluate human-LM interaction, we develop a framework, Human-AI Language-based Interaction Evaluation (H-LINE), that expands non-interactive evaluation along three dimensions, capturing (i) the interactive process, not only the final output; (ii) the first-person subjective experience, not just a third-party assessment; and (iii) notions of preference beyond quality. We then design five tasks ranging from goal-oriented to open-ended to capture different forms of interaction. On four state-of-the-art LMs (three variants of OpenAI's GPT-3 and AI21's J1-Jumbo), we find that non-interactive performance does not always result in better human-LM interaction and that first-person and third-party metrics can diverge, suggesting the importance of examining the nuances of human-LM interaction.

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.

Knowledge-grounded dialogue systems powered by large language models often generate responses that, while fluent, are not attributable to a relevant source of information. Progress towards models that do not exhibit this issue requires evaluation metrics that can quantify its prevalence. To this end, we introduce the Benchmark for Evaluation of Grounded INteraction (BEGIN), comprised of 12k dialogue turns generated by neural dialogue systems trained on three knowledgegrounded dialogue corpora. We collect human annotations assessing the extent to which the models' responses can be attributed to the given background information. We then use BEGIN to analyze eight evaluation metrics. We find that these metrics rely on spurious correlations, do not reliably distinguish attributable abstractive responses from unattributable ones, and perform substantially worse when the knowledge source is longer. Our findings underscore the need for more sophisticated and robust evaluation metrics for knowledge-grounded dialogue. We make BEGIN publicly available at https://github.com/ google/BEGIN-dataset.

近年来，对话系统引起了学术界和工业的重要兴趣。特别是开放式对话系统的纪律，又名聊天，已经获得了很大的势头。然而，困扰研究人员的长期挑战是缺乏有效的自动评估指标，这导致目前研究中的障碍。评估开放式对话模型表现的常见做法涉及对最终部署模型的广泛人类评估，这是时间和成本密集的。此外，最近建立开放式聊天聊天的趋势涉及具有大量社交媒体对话数据的预训练对话模型。但是，社交媒体对话中包含的信息可能是令人反感的和不合适的。不分青红皂白种的使用可能导致不敏感和有毒的生成模型。本文介绍了对话系统技术挑战10（DSTC10）的轨道5获得的数据，基线和结果。

我们提出了一种基于情节知识图（EKG）的新方法，用于评估开放域中的（多模式）对话剂。该图是通过解释对话过程中的原始信号而生成的，并且能够随着时间的推移捕获知识的积累。我们应用对所得图的结构和语义分析，并将这些属性转化为定性措施。我们将这些措施与通常用于对话代理的现有自动和手动评估指标进行比较。我们的结果表明，我们的基于知识的评估为互动和代理人的行为提供了更多的定性见解。

在寻求信息的对话中，用户与代理商进行对话，以提出一系列通常可以不足或过度指定的问题。理想的代理商首先将通过搜索其基本知识来源，然后与用户进行适当互动以解决它，从而确定他们处于这种情况。但是，大多数现有研究都无法或人为地纳入此类代理端计划。在这项工作中，我们介绍了Inscit（发音为Insight），这是一种用于与混合互动相互作用的信息寻求对话的数据集。它包含从805个人类对话中进行的4.7k用户代理转弯，代理商对Wikipedia进行搜索，并要求澄清或提供相关信息以解决用户查询。我们定义了两个子任务，即证据通过识别和响应产生，以及一种新的人类评估协议来评估模型绩效。我们根据对话知识识别和开放域问题的最新模型报告了两个强大的基线的结果。这两种模型都显着不足，并且没有产生连贯和信息丰富的反应，这表明未来的研究有足够的改进空间。

Due to the lack of human resources for mental health support, there is an increasing demand for employing conversational agents for support. Recent work has demonstrated the effectiveness of dialogue models in providing emotional support. As previous studies have demonstrated that seekers' persona is an important factor for effective support, we investigate whether there are benefits to modeling such information in dialogue models for support. In this paper, our empirical analysis verifies that persona has an important impact on emotional support. Therefore, we propose a framework for dynamically inferring and modeling seekers' persona. We first train a model for inferring the seeker's persona from the conversation history. Accordingly, we propose PAL, a model that leverages persona information and, in conjunction with our strategy-based controllable generation method, provides personalized emotional support. Automatic and manual evaluations demonstrate that our proposed model, PAL, achieves state-of-the-art results, outperforming the baselines on the studied benchmark. Our code and data are publicly available at https://github.com/chengjl19/PAL.

以任务为导向的对话系统（TODS）继续升高，因为各种行业发现有效地利用其能力，节省时间和金钱。然而，即使是最先进的TOD尚未达到其全部潜力。TOD通常具有主要设计专注于完成手头的任务，因此任务分辨率的度量应优先考虑。可能会忽略可能指向对话的其他可能指向成功或其他方面的会话质量属性。这可能导致人类和对话系统之间的相互作用，让用户不满意或沮丧。本文探讨了对话系统的评价框架的文献，以及对话系统中的会话质量属性的作用，看起来，如何以及在与对话系统的性能相关的情况下，如何相关。

最先进的对话模型仍然对事实准确性和自我矛盾甚至困难。轶事，他们已被观察到在整个话语中未能维持性质身份;更具体地，可能会涉及其对话者的作用。在这项工作中，我们正规化和量化这种缺陷，并通过人类评估实验表明这确实是一个问题。相比之下，我们展示了专门识别谁在谈话的歧视模型可以表现良好;此外，这些可以用作自动指标。最后，我们评估了各种缓解方法，包括模型架构，培训协议和解码策略的变化。根据人类的注释者，我们最好的车型减少了近65％的误认为是近65％，同时提高了参与度。尽管有这些结果，但我们发现维持性格身份仍然是一个具有挑战性的问题。