我们介绍了Gaudi，Gaudi是一种生成模型，能够捕获可以从移动的相机中沉浸式的复杂和现实3D场景的分布。我们通过一种可扩展而强大的方法解决了这个具有挑战性的问题，我们首先优化了散布辐射场和相机姿势的潜在表示。然后，该潜在表示将学习一个生成模型，该模型可以使3D场景的无条件生成和条件生成。我们的模型概括了以前的作品，该作品通过删除可以在样本中共享相机姿势分布的假设来关注单个对象。我们表明，高迪（Gaudi）在多个数据集的无条件生成设置中获得了最先进的性能，并允许有条件地生成3D场景给定的调理变量，例如稀疏图像观测值或描述场景的文本。

由于具有强大的代表性，变形金刚在包括自然语言处理（NLP），计算机视觉和语音识别在内的广泛应用中越来越受欢迎。但是，利用这种代表性的能力有效地需要大量的数据，强大的正则化或两者兼而有之以减轻过度拟合。最近，基于掩盖的自动编码器的自我监督预处理策略已解锁了变压器的功能，这些策略依赖于直接或从未掩盖的内容对比的掩蔽输入进行重建。这种预训练的策略已在NLP中的BERT模型，Speak2VEC模型中使用，最近在Vision中的MAE模型中，该模型迫使该模型使用自动编码相关的目标来了解输入不同部分中的内容之间的关系。在本文中，我们提出了一种小说但令人惊讶的简单替代内容，以预测内容的位置，而无需为其提供位置信息。这样做需要变压器仅凭内容就可以理解输入不同部分之间的位置关系。这相当于有效的实现，其中借口任务是每个输入令牌所有可能位置之间的分类问题。我们在视觉和语音基准上进行了实验，我们的方法对强有力的监督训练基准进行了改进，并且与现代的无监督/自我监督预审方法相媲美。我们的方法还可以使经过训练的变压器在没有位置嵌入的情况下胜过训练有完整位置信息的训练的变压器。

Power等人报道的\ emph {grokking现象} {power2021grokking}是指一个长期过度拟合之后，似乎突然过渡到完美的概括。在本文中，我们试图通过一系列经验研究来揭示Grokking的基础。具体而言，我们在极端的训练阶段（称为\ emph {slingshot机构）发现了一个优化的异常缺陷自适应优化器。可以通过稳定和不稳定的训练方案之间的循环过渡来测量弹弓机制的突出伪像，并且可以通过最后一层重量的规范的循环行为轻松监测。我们从经验上观察到，在\ cite {power2021grokking}中报道的无明确正规化，几乎完全发生在\ emph {slingshots}的开始时，并且没有它。虽然在更一般的环境中常见且容易复制，但弹弓机制并不遵循我们所知道的任何已知优化理论，并且可以轻松地忽略而无需深入研究。我们的工作表明，在培训的后期阶段，适应性梯度优化器的令人惊讶且有用的归纳偏见，要求对其起源进行修订。

通过提供丰富的训练信号来塑造代理人的潜国空间，建模世界可以使机器人学习受益。然而，在诸如图像之类的高维观察空间上的无约束环境中学习世界模型是具有挑战性的。一个难度来源是存在无关但难以模范的背景干扰，以及不重要的任务相关实体的视觉细节。我们通过学习经常性潜在的动态模型来解决这个问题，该模型对比预测下一次观察。即使使用同时的相机，背景和色调分散，这种简单的模型也会导致令人惊讶的鲁棒机器人控制。我们优于替代品，如双刺激方法，这些方法施加来自未来奖励或未来最佳行为的不同性措施。我们在分散注意力控制套件上获得最先进的结果，是基于像素的机器人控制的具有挑战性的基准。

Reinforcement learning (RL) problems can be challenging without well-shaped rewards. Prior work on provably efficient RL methods generally proposes to address this issue with dedicated exploration strategies. However, another way to tackle this challenge is to reformulate it as a multi-task RL problem, where the task space contains not only the challenging task of interest but also easier tasks that implicitly function as a curriculum. Such a reformulation opens up the possibility of running existing multi-task RL methods as a more efficient alternative to solving a single challenging task from scratch. In this work, we provide a theoretical framework that reformulates a single-task RL problem as a multi-task RL problem defined by a curriculum. Under mild regularity conditions on the curriculum, we show that sequentially solving each task in the multi-task RL problem is more computationally efficient than solving the original single-task problem, without any explicit exploration bonuses or other exploration strategies. We also show that our theoretical insights can be translated into an effective practical learning algorithm that can accelerate curriculum learning on simulated robotic tasks.

In recent years, large amounts of effort have been put into pushing forward the real-world application of dynamic digital human (DDH). However, most current quality assessment research focuses on evaluating static 3D models and usually ignores motion distortions. Therefore, in this paper, we construct a large-scale dynamic digital human quality assessment (DDH-QA) database with diverse motion content as well as multiple distortions to comprehensively study the perceptual quality of DDHs. Both model-based distortion (noise, compression) and motion-based distortion (binding error, motion unnaturalness) are taken into consideration. Ten types of common motion are employed to drive the DDHs and a total of 800 DDHs are generated in the end. Afterward, we render the video sequences of the distorted DDHs as the evaluation media and carry out a well-controlled subjective experiment. Then a benchmark experiment is conducted with the state-of-the-art video quality assessment (VQA) methods and the experimental results show that existing VQA methods are limited in assessing the perceptual loss of DDHs. The database will be made publicly available to facilitate future research.

Data-Free Class Incremental Learning (DFCIL) aims to sequentially learn tasks with access only to data from the current one. DFCIL is of interest because it mitigates concerns about privacy and long-term storage of data, while at the same time alleviating the problem of catastrophic forgetting in incremental learning. In this work, we introduce robust saliency guidance for DFCIL and propose a new framework, which we call RObust Saliency Supervision (ROSS), for mitigating the negative effect of saliency drift. Firstly, we use a teacher-student architecture leveraging low-level tasks to supervise the model with global saliency. We also apply boundary-guided saliency to protect it from drifting across object boundaries at intermediate layers. Finally, we introduce a module for injecting and recovering saliency noise to increase robustness of saliency preservation. Our experiments demonstrate that our method can retain better saliency maps across tasks and achieve state-of-the-art results on the CIFAR-100, Tiny-ImageNet and ImageNet-Subset DFCIL benchmarks. Code will be made publicly available.

Vision Transformers convert images to sequences by slicing them into patches. The size of these patches controls a speed/accuracy tradeoff, with smaller patches leading to higher accuracy at greater computational cost, but changing the patch size typically requires retraining the model. In this paper, we demonstrate that simply randomizing the patch size at training time leads to a single set of weights that performs well across a wide range of patch sizes, making it possible to tailor the model to different compute budgets at deployment time. We extensively evaluate the resulting model, which we call FlexiViT, on a wide range of tasks, including classification, image-text retrieval, open-world detection, panoptic segmentation, and semantic segmentation, concluding that it usually matches, and sometimes outperforms, standard ViT models trained at a single patch size in an otherwise identical setup. Hence, FlexiViT training is a simple drop-in improvement for ViT that makes it easy to add compute-adaptive capabilities to most models relying on a ViT backbone architecture. Code and pre-trained models are available at https://github.com/google-research/big_vision

Semantic segmentation usually benefits from global contexts, fine localisation information, multi-scale features, etc. To advance Transformer-based segmenters with these aspects, we present a simple yet powerful semantic segmentation architecture, termed as IncepFormer. IncepFormer has two critical contributions as following. First, it introduces a novel pyramid structured Transformer encoder which harvests global context and fine localisation features simultaneously. These features are concatenated and fed into a convolution layer for final per-pixel prediction. Second, IncepFormer integrates an Inception-like architecture with depth-wise convolutions, and a light-weight feed-forward module in each self-attention layer, efficiently obtaining rich local multi-scale object features. Extensive experiments on five benchmarks show that our IncepFormer is superior to state-of-the-art methods in both accuracy and speed, e.g., 1) our IncepFormer-S achieves 47.7% mIoU on ADE20K which outperforms the existing best method by 1% while only costs half parameters and fewer FLOPs. 2) Our IncepFormer-B finally achieves 82.0% mIoU on Cityscapes dataset with 39.6M parameters. Code is available:github.com/shendu0321/IncepFormer.

A common scenario of Multilingual Neural Machine Translation (MNMT) is that each translation task arrives in a sequential manner, and the training data of previous tasks is unavailable. In this scenario, the current methods suffer heavily from catastrophic forgetting (CF). To alleviate the CF, we investigate knowledge distillation based life-long learning methods. Specifically, in one-tomany scenario, we propose a multilingual distillation method to make the new model (student) jointly learn multilingual output from old model (teacher) and new task. In many-to one scenario, we find that direct distillation faces the extreme partial distillation problem, and we propose two different methods to address it: pseudo input distillation and reverse teacher distillation. The experimental results on twelve translation tasks show that the proposed methods can better consolidate the previous knowledge and sharply alleviate the CF.