Fusion-in-Decoder (FiD) is a powerful retrieval-augmented language model that sets the state-of-the-art on many knowledge-intensive NLP tasks. However, FiD suffers from very expensive inference. We show that the majority of inference time results from memory bandwidth constraints in the decoder, and propose two simple changes to the FiD architecture to speed up inference by 7x. The faster decoder inference then allows for a much larger decoder. We denote FiD with the above modifications as FiDO, and show that it strongly improves performance over existing FiD models for a wide range of inference budgets. For example, FiDO-Large-XXL performs faster inference than FiD-Base and achieves better performance than FiD-Large.

在该职位论文中，我们提出了一种新方法，以基于问题的产生和实体链接来生成文本的知识库（KB）。我们认为，所提出的KB类型具有传统符号KB的许多关键优势：尤其是由小型模块化组件组成，可以在组合上合并以回答复杂的查询，包括涉及“多跳跃”的关系查询和查询。“推论。但是，与传统的KB不同，该信息商店与常见的用户信息需求相符。

最近的证据指出了高性能跨度预测模型的脆弱的动机，我们将注意力指向多种选择阅读理解。特别是，这项工作介绍了一种新的方法，用于通过重量全球正常化改进答案选择，通过对文档的一部分的预测的加权全球化。我们表明，将我们的方法应用于适用于答案选择的跨度预测模型，有助于从叙述问题的长摘要进行模型性能，这是一个充满挑战的阅读理解数据集，具有答案选择任务，我们强烈提高任务基线性能+36.2平均互酷等级。

Large language models (LLMs) have been shown to be able to perform new tasks based on a few demonstrations or natural language instructions. While these capabilities have led to widespread adoption, most LLMs are developed by resource-rich organizations and are frequently kept from the public. As a step towards democratizing this powerful technology, we present BLOOM, a 176B-parameter open-access language model designed and built thanks to a collaboration of hundreds of researchers. BLOOM is a decoder-only Transformer language model that was trained on the ROOTS corpus, a dataset comprising hundreds of sources in 46 natural and 13 programming languages (59 in total). We find that BLOOM achieves competitive performance on a wide variety of benchmarks, with stronger results after undergoing multitask prompted finetuning. To facilitate future research and applications using LLMs, we publicly release our models and code under the Responsible AI License.

Policies produced by deep reinforcement learning are typically characterised by their learning curves, but they remain poorly understood in many other respects. ReLU-based policies result in a partitioning of the input space into piecewise linear regions. We seek to understand how observed region counts and their densities evolve during deep reinforcement learning using empirical results that span a range of continuous control tasks and policy network dimensions. Intuitively, we may expect that during training, the region density increases in the areas that are frequently visited by the policy, thereby affording fine-grained control. We use recent theoretical and empirical results for the linear regions induced by neural networks in supervised learning settings for grounding and comparison of our results. Empirically, we find that the region density increases only moderately throughout training, as measured along fixed trajectories coming from the final policy. However, the trajectories themselves also increase in length during training, and thus the region densities decrease as seen from the perspective of the current trajectory. Our findings suggest that the complexity of deep reinforcement learning policies does not principally emerge from a significant growth in the complexity of functions observed on-and-around trajectories of the policy.

Reinforcement Learning (RL) has seen many recent successes for quadruped robot control. The imitation of reference motions provides a simple and powerful prior for guiding solutions towards desired solutions without the need for meticulous reward design. While much work uses motion capture data or hand-crafted trajectories as the reference motion, relatively little work has explored the use of reference motions coming from model-based trajectory optimization. In this work, we investigate several design considerations that arise with such a framework, as demonstrated through four dynamic behaviours: trot, front hop, 180 backflip, and biped stepping. These are trained in simulation and transferred to a physical Solo 8 quadruped robot without further adaptation. In particular, we explore the space of feed-forward designs afforded by the trajectory optimizer to understand its impact on RL learning efficiency and sim-to-real transfer. These findings contribute to the long standing goal of producing robot controllers that combine the interpretability and precision of model-based optimization with the robustness that model-free RL-based controllers offer.

癌症护理中的治疗决策受到随机对照试验（RCT）的治疗效应估计的指导。 RCT估计在某个人群中，一种治疗与另一种治疗的平均效应。但是，治疗可能对人群中的每个患者都不同样有效。了解针对特定患者和肿瘤特征量身定制的治疗的有效性将实现个性化的治疗决策。通过平均RCT中不同患者亚组的结果来获得量身定制的治疗效果，需要大量的患者在所有相关亚组中具有足够的统计能力，以实现所有可能的治疗。美国癌症联合委员会（AJCC）建议研究人员开发结果预测模型（OPMS），以实现个性化治疗决策。 OPM有时称为风险模型或预后模型，使用患者和肿瘤特征来预测患者的结局，例如总体生存。假设这些预测对于使用“只有在OPM预测患者具有高复发风险的情况下开出化学疗法的规则”之类的规则，对治疗决策有用。 AJCC认识到可靠预测的重要性，发布了OPM的清单，以确保设计OPM设计的患者群体的可靠OPM预测准确性。但是，准确的结果预测并不意味着这些预测会产生良好的治疗决策。从这个角度来看，我们表明OPM依靠固定的治疗政策，这意味着被发现可以准确预测验证研究结果的OPM在用于治疗决策的情况下仍会导致患者伤害。然后，我们提供有关如何开发对个性化治疗决策有用的模型以及如何评估模型是否具有决策价值的指导。

从任意堕落状态中起床是一种基本的人类技能。现有的学习这种技能的方法通常会产生高度动态和不稳定的起床动作，这不像人类的起床策略，或者基于跟踪记录的人类起床运动。在本文中，我们提出了一种使用强化学习的分阶段方法，而无需求助于运动捕获数据。该方法首先利用了强大的字符模型，从而有助于发现解决方案模式。然后，第二阶段学会了调整控制策略，以逐步与角色的较弱版本一起使用。最后，第三阶段学习控制政策，这些政策可以以较慢的速度重现较弱的起床动作。我们表明，在多个运行中，该方法可以发现各种各样的起床策略，并以各种速度执行它们。结果通常会产生采用最终站立策略的策略，这些策略是从所有初始状态中看到的恢复动作所共有的。但是，我们还发现了对俯卧和仰卧初始堕落状态的不同策略的政策。学识渊博的起床控制策略通常具有明显的静态稳定性，即，在起床运动过程中，它们可以在各个点停下来。我们进一步测试了新的限制场景的方法，例如在演员表中有一条腿和手臂。

深度神经网络在预测质量方面表现出巨大的成功，而可靠且稳健的不确定性估计仍然是一个挑战。预测性不确定性补充模型预测，并实现了下游任务的改进功能，包括嵌入式和移动应用，例如虚拟现实，增强现实，传感器融合和感知。这些应用程序通常需要复杂性的妥协，以获得由于内存非常有限和计算资源而导致的不确定性估计。我们通过使用Axolotl框架构建Monte Carlo辍学（MCDO）模型来解决这个问题;具体而言，我们多样化采样的子网，利用辍学模式，并使用分支技术来提高预测性能，同时保持快速计算。我们在使用CIFAR10 DataSet上进行（1）多级分类任务的实验，（2）更复杂的人体分段任务。我们的结果表明我们的方法通过接近深度集成预测质量和不确定性估算来达到效果，同时仍在实现资源限制的移动平台的推断。

尽管近期因因果推断领域的进展，迄今为止没有关于从观察数据的收集治疗效应估算的方法。对临床实践的结果是，当缺乏随机试验的结果时，没有指导在真实情景中似乎有效的指导。本文提出了一种务实的方法，以获得从观察性研究的治疗效果的初步但稳健地估算，为前线临床医生提供对其治疗策略的信心程度。我们的研究设计适用于一个公开问题，估算Covid-19密集护理患者的拳击机动的治疗效果。