Recent work has shown that fine-tuning large pre-trained language models on a collection of tasks described via instructions, a.k.a. instruction-tuning, improves their zero and few-shot generalization to unseen tasks. However, there is a limited understanding of the performance trade-offs of different decisions made during the instruction-tuning process. These decisions include the scale and diversity of the instruction-tuning benchmark, different task sampling strategies, fine-tuning with and without demonstrations, training using specialized datasets for reasoning and dialogue, and finally, the fine-tuning objectives themselves. In this paper, we characterize the effect of instruction-tuning decisions on downstream task performance when scaling both model and benchmark sizes. To this end, we create OPT-IML Bench: a large benchmark for Instruction Meta-Learning (IML) of 2000 NLP tasks consolidated into task categories from 8 existing benchmarks, and prepare an evaluation framework to measure three types of model generalizations: to tasks from fully held-out categories, to held-out tasks from seen categories, and to held-out instances from seen tasks. Through the lens of this framework, we first present insights about instruction-tuning decisions as applied to OPT-30B and further exploit these insights to train OPT-IML 30B and 175B, which are instruction-tuned versions of OPT. OPT-IML demonstrates all three generalization abilities at both scales on four different evaluation benchmarks with diverse tasks and input formats -- PromptSource, FLAN, Super-NaturalInstructions, and UnifiedSKG. Not only does it significantly outperform OPT on all benchmarks but is also highly competitive with existing models fine-tuned on each specific benchmark. We release OPT-IML at both scales, together with the OPT-IML Bench evaluation framework.

Recent progress on vision-language foundation models have brought significant advancement to building general-purpose robots. By using the pre-trained models to encode the scene and instructions as inputs for decision making, the instruction-conditioned policy can generalize across different objects and tasks. While this is encouraging, the policy still fails in most cases given an unseen task or environment. To adapt the policy to unseen tasks and environments, we explore a new paradigm on leveraging the pre-trained foundation models with Self-PLAY and Self-Describe (SPLAYD). When deploying the trained policy to a new task or a new environment, we first let the policy self-play with randomly generated instructions to record the demonstrations. While the execution could be wrong, we can use the pre-trained foundation models to accurately self-describe (i.e., re-label or classify) the demonstrations. This automatically provides new pairs of demonstration-instruction data for policy fine-tuning. We evaluate our method on a broad range of experiments with the focus on generalization on unseen objects, unseen tasks, unseen environments, and sim-to-real transfer. We show SPLAYD improves baselines by a large margin in all cases. Our project page is available at https://geyuying.github.io/SPLAYD/

This paper proposes a deep recurrent Rotation Averaging Graph Optimizer (RAGO) for Multiple Rotation Averaging (MRA). Conventional optimization-based methods usually fail to produce accurate results due to corrupted and noisy relative measurements. Recent learning-based approaches regard MRA as a regression problem, while these methods are sensitive to initialization due to the gauge freedom problem. To handle these problems, we propose a learnable iterative graph optimizer minimizing a gauge-invariant cost function with an edge rectification strategy to mitigate the effect of inaccurate measurements. Our graph optimizer iteratively refines the global camera rotations by minimizing each node's single rotation objective function. Besides, our approach iteratively rectifies relative rotations to make them more consistent with the current camera orientations and observed relative rotations. Furthermore, we employ a gated recurrent unit to improve the result by tracing the temporal information of the cost graph. Our framework is a real-time learning-to-optimize rotation averaging graph optimizer with a tiny size deployed for real-world applications. RAGO outperforms previous traditional and deep methods on real-world and synthetic datasets. The code is available at https://github.com/sfu-gruvi-3dv/RAGO

Stochastic gradients closely relate to both optimization and generalization of deep neural networks (DNNs). Some works attempted to explain the success of stochastic optimization for deep learning by the arguably heavy-tail properties of gradient noise, while other works presented theoretical and empirical evidence against the heavy-tail hypothesis on gradient noise. Unfortunately, formal statistical tests for analyzing the structure and heavy tails of stochastic gradients in deep learning are still under-explored. In this paper, we mainly make two contributions. First, we conduct formal statistical tests on the distribution of stochastic gradients and gradient noise across both parameters and iterations. Our statistical tests reveal that dimension-wise gradients usually exhibit power-law heavy tails, while iteration-wise gradients and stochastic gradient noise caused by minibatch training usually do not exhibit power-law heavy tails. Second, we further discover that the covariance spectra of stochastic gradients have the power-law structures in deep learning. While previous papers believed that the anisotropic structure of stochastic gradients matters to deep learning, they did not expect the gradient covariance can have such an elegant mathematical structure. Our work challenges the existing belief and provides novel insights on the structure of stochastic gradients in deep learning.

Automated detecting lung infections from computed tomography (CT) data plays an important role for combating COVID-19. However, there are still some challenges for developing AI system. 1) Most current COVID-19 infection segmentation methods mainly relied on 2D CT images, which lack 3D sequential constraint. 2) Existing 3D CT segmentation methods focus on single-scale representations, which do not achieve the multiple level receptive field sizes on 3D volume. 3) The emergent breaking out of COVID-19 makes it hard to annotate sufficient CT volumes for training deep model. To address these issues, we first build a multiple dimensional-attention convolutional neural network (MDA-CNN) to aggregate multi-scale information along different dimension of input feature maps and impose supervision on multiple predictions from different CNN layers. Second, we assign this MDA-CNN as a basic network into a novel dual multi-scale mean teacher network (DM${^2}$T-Net) for semi-supervised COVID-19 lung infection segmentation on CT volumes by leveraging unlabeled data and exploring the multi-scale information. Our DM${^2}$T-Net encourages multiple predictions at different CNN layers from the student and teacher networks to be consistent for computing a multi-scale consistency loss on unlabeled data, which is then added to the supervised loss on the labeled data from multiple predictions of MDA-CNN. Third, we collect two COVID-19 segmentation datasets to evaluate our method. The experimental results show that our network consistently outperforms the compared state-of-the-art methods.

End-to-end autonomous driving provides a feasible way to automatically maximize overall driving system performance by directly mapping the raw pixels from a front-facing camera to control signals. Recent advanced methods construct a latent world model to map the high dimensional observations into compact latent space. However, the latent states embedded by the world model proposed in previous works may contain a large amount of task-irrelevant information, resulting in low sampling efficiency and poor robustness to input perturbations. Meanwhile, the training data distribution is usually unbalanced, and the learned policy is hard to cope with the corner cases during the driving process. To solve the above challenges, we present a semantic masked recurrent world model (SEM2), which introduces a latent filter to extract key task-relevant features and reconstruct a semantic mask via the filtered features, and is trained with a multi-source data sampler, which aggregates common data and multiple corner case data in a single batch, to balance the data distribution. Extensive experiments on CARLA show that our method outperforms the state-of-the-art approaches in terms of sample efficiency and robustness to input permutations.

在训练之前对神经网络进行修剪不仅会压缩原始模型，还可以加速其具有实质性应用值的网络培训阶段。当前的工作着重于细粒修剪，该修剪使用指标来计算重量筛查的重量评分，并从初始的单阶修剪到迭代修剪。通过这些作品，我们认为可以将网络修剪总结为权重的表达力传递过程，其中保留权重将从被删除的力量中占据表达力，以维持原始网络的性能。为了实现最佳的表达力调度，我们在训练名为神经网络Pannaging之前提出了一种修剪计划，该方案通过多指数和多进程步骤指导表达力转移，并设计一种基于强化学习以自动化过程的平移代理。实验结果表明，平平在训练方法之前的性能优于各种可用的修剪。

现有的广告点击率（CTR）预测模型主要取决于行为ID功能，这些功能是根据历史用户AD交互所学习的。然而，依赖历史用户行为的行为ID功能是不可行的，可以在没有以前与用户互动的情况下描述新广告。为了克服对新广告建模的行为ID特征的局限性，我们利用广告中的视觉内容来提高CTR预测模型的性能。具体来说，我们根据其视觉内容将每个广告映射到一组视觉ID中。这些视觉ID进一步用于生成可视觉嵌入，以增强CTR预测模型。我们将视觉ID的学习分为有监督的量化问题。由于缺乏广告中商业图像的类标签，因此我们利用图像文本描述作为监督，以优化图像提取器以生成有效的视觉ID。同时，由于硬量化是不可差异的，因此我们软化量化操作以使其支持端到端网络培训。将每个图像映射到视觉ID之后，我们根据过去积累的历史用户AD交互学习每个视觉ID的嵌入。由于视觉ID嵌入仅取决于视觉内容，因此它概括为新广告。同时，嵌入视觉ID补充了AD行为ID嵌入。因此，它可以大大提高CTR预测模型的性能，以前依赖于积累了丰富用户行为的新广告和广告的行为ID功能。将视觉ID嵌入在BAIDU在线广告的CTR预测模型中后，AD的平均CTR提高了1.46％，总费用增加了1.10％。

通信技术的进步和智能手机的普及促进了视频广告的蓬勃发展。百度是世界领先的搜索引擎公司之一，每天收到数十亿个搜索查询。如何将视频广告与用户搜索配对是百度视频广告的核心任务。由于模态差距，比传统查询对象检索和图像到图像搜索更具挑战性的查询性检索更具挑战性。传统上，查询到视频检索是通过查询到标题检索来解决的，当瓷砖的质量不高时，这是不可靠的。近年来，随着计算机视觉和自然语言处理的快速进展，基于内容的搜索方法变得有望在查询到视频检索中。受益于大规模数据集的预处理，一些基于跨模式关注的Visionbert方法在许多视觉语言任务中不仅在学术界而且在行业中都取得了出色的表现。然而，跨模式关注的昂贵计算成本使得在工业应用中进行大规模搜索是不切实际的。在这项工作中，我们提出了一个基于树的组合注意网络（TCAN），该网络最近在百度的动态视频广告平台上推出。它提供了一种实用的解决方案，可以在大规模查询到视频搜索中部署大量的跨模式关注。在启动基于树的组合注意网络之后，点击率提高了2.29 \％，转化率提高了2.63 \％。

在鸟眼中学习强大的表现（BEV），以进行感知任务，这是趋势和吸引行业和学术界的广泛关注。大多数自动驾驶算法的常规方法在正面或透视视图中执行检测，细分，跟踪等。随着传感器配置变得越来越复杂，从不同的传感器中集成了多源信息，并在统一视图中代表功能至关重要。 BEV感知继承了几个优势，因为代表BEV中的周围场景是直观和融合友好的。对于BEV中的代表对象，对于随后的模块，如计划和/或控制是最可取的。 BEV感知的核心问题在于（a）如何通过从透视视图到BEV来通过视图转换来重建丢失的3D信息； （b）如何在BEV网格中获取地面真理注释； （c）如何制定管道以合并来自不同来源和视图的特征； （d）如何适应和概括算法作为传感器配置在不同情况下各不相同。在这项调查中，我们回顾了有关BEV感知的最新工作，并对不同解决方案进行了深入的分析。此外，还描述了该行业的BEV方法的几种系统设计。此外，我们推出了一套完整的实用指南，以提高BEV感知任务的性能，包括相机，激光雷达和融合输入。最后，我们指出了该领域的未来研究指示。我们希望该报告能阐明社区，并鼓励对BEV感知的更多研究。我们保留一个活跃的存储库来收集最新的工作，并在https://github.com/openperceptionx/bevperception-survey-recipe上提供一包技巧的工具箱。