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.

许多NLP任务受益于使用通常具有超过1000亿参数的大语言模型（LLM）。随着Bloom-176b和Opt-175B的发布，每个人都可以下载该规模的预估计型号。尽管如此，使用这些模型仍需要许多研究人员无法获得高端硬件。在某些情况下，LLM可以通过RAM卸载或托管API更实惠。但是，这些技术具有先天的局限性：对于交互推理而言，卸载太慢，而API的灵活性不足以进行研究。在这项工作中，我们通过加入信任处理客户数据的多个政党的资源来提出花瓣$ - $ $用于推理和微调大型模型的系统。我们证明，这种策略的表现极大地超过了非常大型型号的卸载，以每秒约1美元的价格$ \ $ \ $ \ $ \ $ \ $ \ $ \ $ \ $ 1。与大多数推理API不同，花瓣还本地揭示了服务模型的隐藏状态，从而使其用户可以根据有效的微调方法训练和共享自定义模型扩展。

培训最先进模型所需的基础设施变得过于昂贵，这使得培训此类模型仅适用于大型公司和机构。最近的工作提出了几种协作培训此类模型的方法，即通过将许多独立方的硬件汇总在一起，并通过Internet培训共享模型。在此演示中，我们合作培训了类似于Openai Dall-E的文本到图像变压器。我们邀请观众加入正在进行的训练运行，向他们展示有关如何使用可用硬件贡献的说明。我们解释了如何应对与此类训练运行相关的工程挑战（缓慢的沟通，有限的内存，设备之间的性能不均和安全问题），并讨论了观众如何设置协作培训。最后，我们表明所得模型在许多提示上生成了合理质量的图像。

致辞推理是自然语言处理中的关键问题之一，但标记数据的相对稀缺缺少英语以外的语言的进度。预先磨削的交叉模型是一种强大的语言不可知论者的来源，但它们的固有推理能力仍在积极研究。在这项工作中，我们设计了一种简单的方法来推理，将线性分类器列举为具有多针关注的重量。为了评估这种方法，我们通过在标准化管道内的先前工作中处理多种数据集来创建多语言WinoGrad模式语料库，并在样品外性能方面测量交叉语言泛化能力。该方法在近期监督和无人监督推理的最近监督和无监督的方法中表现得很竞争，即使在以零拍摄方式应用于其他语言。此外，我们证明大多数性能由所有研究的语言的相同小的注意头给出，这提供了在多语言编码器中的普遍推理能力的证据。

许多深度学习领域都受益于使用越来越大的神经网络接受公共数据训练的培训，就像预先训练的NLP和计算机视觉模型一样。培训此类模型需要大量的计算资源（例如，HPC群集），而小型研究小组和独立研究人员则无法使用。解决问题的一种方法是，几个较小的小组将其计算资源汇总在一起并训练一种使所有参与者受益的模型。不幸的是，在这种情况下，任何参与者都可以通过故意或错误地发送错误的更新来危害整个培训。在此类同龄人的情况下进行培训需要具有拜占庭公差的专门分布式培训算法。这些算法通常通过引入冗余通信或通过受信任的服务器传递所有更新来牺牲效率，从而使它们无法应用于大规模深度学习，在该大规模深度学习中，模型可以具有数十亿个参数。在这项工作中，我们提出了一种新的协议，用于强调沟通效率的安全（容忍）分散培训。

现代深度学习应用程序需要越来越多地计算培训最先进的模型。为了解决这一需求，大型企业和机构使用专用的高性能计算集群，其建筑和维护既昂贵又远远超出大多数组织的预算。结果，一些研究方向成为几个大型工业甚至更少的学术作用者的独家领域。为了减轻这种差异，较小的团体可以汇集他们的计算资源并运行有利于所有参与者的协作实验。这种范式称为网格或志愿者计算，在众多科学领域看到了成功的应用。然而，由于高延迟，不对称带宽以及志愿者计算独特的几个挑战，使用这种用于机器学习的方法是困难的。在这项工作中，我们仔细分析了这些约束，并提出了一种专门用于协作培训的新型算法框架。我们展示了我们在现实条件下的SWAV和Albert预先预价的方法的有效性，并在成本的一小部分中实现了与传统设置相当的性能。最后，我们提供了一份成功的协作语言模型预先追溯的详细报告，有40名参与者。

使用多个计算节点通常可以加速在大型数据集上的深度神经网络。这种方法称为分布式训练，可以通过专门的消息传递协议，例如环形全部减少。但是，以比例运行这些协议需要可靠的高速网络，其仅在专用集群中可用。相比之下，许多现实世界应用程序，例如联合学习和基于云的分布式训练，在具有不稳定的网络带宽的不可靠的设备上运行。因此，这些应用程序仅限于使用参数服务器或基于Gossip的平均协议。在这项工作中，我们通过提出MOSHPIT全部减少的迭代平均协议来提升该限制，该协议指数地收敛于全局平均值。我们展示了我们对具有强烈理论保证的分布式优化方案的效率。该实验显示了与使用抢占从头开始训练的竞争性八卦的策略和1.5倍的加速，显示了1.3倍的Imagenet培训的加速。

With the advent of Neural Style Transfer (NST), stylizing an image has become quite popular. A convenient way for extending stylization techniques to videos is by applying them on a per-frame basis. However, such per-frame application usually lacks temporal-consistency expressed by undesirable flickering artifacts. Most of the existing approaches for enforcing temporal-consistency suffers from one or more of the following drawbacks. They (1) are only suitable for a limited range of stylization techniques, (2) can only be applied in an offline fashion requiring the complete video as input, (3) cannot provide consistency for the task of stylization, or (4) do not provide interactive consistency-control. Note that existing consistent video-filtering approaches aim to completely remove flickering artifacts and thus do not respect any specific consistency-control aspect. For stylization tasks, however, consistency-control is an essential requirement where a certain amount of flickering can add to the artistic look and feel. Moreover, making this control interactive is paramount from a usability perspective. To achieve the above requirements, we propose an approach that can stylize video streams while providing interactive consistency-control. Apart from stylization, our approach also supports various other image processing filters. For achieving interactive performance, we develop a lite optical-flow network that operates at 80 Frames per second (FPS) on desktop systems with sufficient accuracy. We show that the final consistent video-output using our flow network is comparable to that being obtained using state-of-the-art optical-flow network. Further, we employ an adaptive combination of local and global consistent features and enable interactive selection between the two. By objective and subjective evaluation, we show that our method is superior to state-of-the-art approaches.

In this work, we address the problem of unsupervised moving object segmentation (MOS) in 4D LiDAR data recorded from a stationary sensor, where no ground truth annotations are involved. Deep learning-based state-of-the-art methods for LiDAR MOS strongly depend on annotated ground truth data, which is expensive to obtain and scarce in existence. To close this gap in the stationary setting, we propose a novel 4D LiDAR representation based on multivariate time series that relaxes the problem of unsupervised MOS to a time series clustering problem. More specifically, we propose modeling the change in occupancy of a voxel by a multivariate occupancy time series (MOTS), which captures spatio-temporal occupancy changes on the voxel level and its surrounding neighborhood. To perform unsupervised MOS, we train a neural network in a self-supervised manner to encode MOTS into voxel-level feature representations, which can be partitioned by a clustering algorithm into moving or stationary. Experiments on stationary scenes from the Raw KITTI dataset show that our fully unsupervised approach achieves performance that is comparable to that of supervised state-of-the-art approaches.

Implicit Neural Representations (INR) have recently shown to be powerful tool for high-quality video compression. However, existing works are limiting as they do not explicitly exploit the temporal redundancy in videos, leading to a long encoding time. Additionally, these methods have fixed architectures which do not scale to longer videos or higher resolutions. To address these issues, we propose NIRVANA, which treats videos as groups of frames and fits separate networks to each group performing patch-wise prediction. This design shares computation within each group, in the spatial and temporal dimensions, resulting in reduced encoding time of the video. The video representation is modeled autoregressively, with networks fit on a current group initialized using weights from the previous group's model. To further enhance efficiency, we perform quantization of the network parameters during training, requiring no post-hoc pruning or quantization. When compared with previous works on the benchmark UVG dataset, NIRVANA improves encoding quality from 37.36 to 37.70 (in terms of PSNR) and the encoding speed by 12X, while maintaining the same compression rate. In contrast to prior video INR works which struggle with larger resolution and longer videos, we show that our algorithm is highly flexible and scales naturally due to its patch-wise and autoregressive designs. Moreover, our method achieves variable bitrate compression by adapting to videos with varying inter-frame motion. NIRVANA achieves 6X decoding speed and scales well with more GPUs, making it practical for various deployment scenarios.