基础模型正在成为主要的深度学习技术。由于模型参数和训练数据集的大规模，预处理基础模型始终耗时。除了计算密集型外，培训过程还非常密集和沟通密集。这些功能使得需要应用3D并行性，该平行性整合数据并行性，管道模型并行性和张量模型并行性，以实现高训练效率。为了实现这一目标，开发了一些自定义软件框架，例如Megatron-LM和DeepSpeed。但是，当前的3D平行框架仍然符合两个问题：i）它们对模型开发人员不透明，这些开发人员需要手动修改模型以并行化培训。 ii）它们对计算，GPU存储器和网络带宽的利用不足。我们提出了Merak，这是一个自动化的3D并行性深度学习培训框架，并具有高度资源利用。 Merak会自动使用自动模型分区仪部署，该分区仪在模型的代理表示上使用图形sharding算法。 Merak还提出了非侵入性的API，用于通过最小的代码修改来扩展基础模型培训。此外，我们在Merak设计了高性能的3D平行运行时引擎。它使用多种技术来利用可用的培训资源，包括移动的关键路径管道时间表，该计划带来了更高的计算利用率，阶段感知的重新计算，可利用空闲工作者的记忆以及子额定张量的模型并行性，这些模型并联与通信和计算重叠。 64 GPU的实验显示，Merak可以加快在最新的3D平行性框架上，具有1.5、2.5、8.3和20亿的模型框架，最高可达1.42x，1.39x，1.43x和1.61 x分别。

分布式数据并行训练已被广泛用于深神经网络（DNN）模型。尽管当前的深度学习（DL）框架对于图像分类模型（例如图像分类模型）的密集模型很好地扩展了，但我们发现这些DL框架对于具有高度稀疏嵌入表的稀疏模型（NLP）模型（NLP）模型（NLP）模型具有相对较低的可扩展性。大多数现有作品忽略了模型参数的稀疏性，因此遭受了重要但不必要的沟通开销。在本文中，我们提出了Ablace，这是一个有效的沟通框架，以加快稀疏模型分布式培训的通信。 Embrace引入了稀疏感知的混合通信，将AlltoAll和模型并行置于数据并行训练中，以减少高度稀疏参数的交流开销。为了有效地重叠稀疏的通信与后向前和前向计算，采用进一步设计的2D通信调度方法，该方法优化了模型计算过程，放松嵌入式的依赖性，并计划以优先级的排队来安排每个嵌入行的稀疏通信。我们已经基于Pytorch和Horovod实施了Embrace的原型，并通过四个代表性的NLP模型进行了全面的评估。实验结果表明，与最先进的分布式训练基线相比，Embrace的速度高达2.41倍。

Non-line-of-sight (NLOS) imaging aims to reconstruct the three-dimensional hidden scenes from the data measured in the line-of-sight, which uses photon time-of-flight information encoded in light after multiple diffuse reflections. The under-sampled scanning data can facilitate fast imaging. However, the resulting reconstruction problem becomes a serious ill-posed inverse problem, the solution of which is of high possibility to be degraded due to noises and distortions. In this paper, we propose two novel NLOS reconstruction models based on curvature regularization, i.e., the object-domain curvature regularization model and the dual (i.e., signal and object)-domain curvature regularization model. Fast numerical optimization algorithms are developed relying on the alternating direction method of multipliers (ADMM) with the backtracking stepsize rule, which are further accelerated by GPU implementation. We evaluate the proposed algorithms on both synthetic and real datasets, which achieve state-of-the-art performance, especially in the compressed sensing setting. All our codes and data are available at https://github.com/Duanlab123/CurvNLOS.

In this paper, we target at the problem of learning a generalizable dynamic radiance field from monocular videos. Different from most existing NeRF methods that are based on multiple views, monocular videos only contain one view at each timestamp, thereby suffering from ambiguity along the view direction in estimating point features and scene flows. Previous studies such as DynNeRF disambiguate point features by positional encoding, which is not transferable and severely limits the generalization ability. As a result, these methods have to train one independent model for each scene and suffer from heavy computational costs when applying to increasing monocular videos in real-world applications. To address this, We propose MonoNeRF to simultaneously learn point features and scene flows with point trajectory and feature correspondence constraints across frames. More specifically, we learn an implicit velocity field to estimate point trajectory from temporal features with Neural ODE, which is followed by a flow-based feature aggregation module to obtain spatial features along the point trajectory. We jointly optimize temporal and spatial features by training the network in an end-to-end manner. Experiments show that our MonoNeRF is able to learn from multiple scenes and support new applications such as scene editing, unseen frame synthesis, and fast novel scene adaptation.

In this paper, we propose a large-scale language pre-training for text GENeration using dIffusion modEl, which is named GENIE. GENIE is a pre-training sequence-to-sequence text generation model which combines Transformer and diffusion. The diffusion model accepts the latent information from the encoder, which is used to guide the denoising of the current time step. After multiple such denoise iterations, the diffusion model can restore the Gaussian noise to the diverse output text which is controlled by the input text. Moreover, such architecture design also allows us to adopt large scale pre-training on the GENIE. We propose a novel pre-training method named continuous paragraph denoise based on the characteristics of the diffusion model. Extensive experiments on the XSum, CNN/DailyMail, and Gigaword benchmarks shows that GENIE can achieves comparable performance with various strong baselines, especially after pre-training, the generation quality of GENIE is greatly improved. We have also conduct a lot of experiments on the generation diversity and parameter impact of GENIE. The code for GENIE will be made publicly available.

Structured tabular data exist across nearly all fields. Reasoning task over these data aims to answer questions or determine the truthiness of hypothesis sentences by understanding the semantic meaning of a table. While previous works have devoted significant efforts to the tabular reasoning task, they always assume there are sufficient labeled data. However, constructing reasoning samples over tables (and related text) is labor-intensive, especially when the reasoning process is complex. When labeled data is insufficient, the performance of models will suffer an unendurable decline. In this paper, we propose a unified framework for unsupervised complex tabular reasoning (UCTR), which generates sufficient and diverse synthetic data with complex logic for tabular reasoning tasks, assuming no human-annotated data at all. We first utilize a random sampling strategy to collect diverse programs of different types and execute them on tables based on a "Program-Executor" module. To bridge the gap between the programs and natural language sentences, we design a powerful "NL-Generator" module to generate natural language sentences with complex logic from these programs. Since a table often occurs with its surrounding texts, we further propose novel "Table-to-Text" and "Text-to-Table" operators to handle joint table-text reasoning scenarios. This way, we can adequately exploit the unlabeled table resources to obtain a well-performed reasoning model under an unsupervised setting. Our experiments cover different tasks (question answering and fact verification) and different domains (general and specific), showing that our unsupervised methods can achieve at most 93% performance compared to supervised models. We also find that it can substantially boost the supervised performance in low-resourced domains as a data augmentation technique. Our code is available at https://github.com/leezythu/UCTR.

Making sense of multiple modalities can yield a more comprehensive description of real-world phenomena. However, learning the co-representation of diverse modalities is still a long-standing endeavor in emerging machine learning applications and research. Previous generative approaches for multimodal input approximate a joint-modality posterior by uni-modality posteriors as product-of-experts (PoE) or mixture-of-experts (MoE). We argue that these approximations lead to a defective bound for the optimization process and loss of semantic connection among modalities. This paper presents a novel variational method on sets called the Set Multimodal VAE (SMVAE) for learning a multimodal latent space while handling the missing modality problem. By modeling the joint-modality posterior distribution directly, the proposed SMVAE learns to exchange information between multiple modalities and compensate for the drawbacks caused by factorization. In public datasets of various domains, the experimental results demonstrate that the proposed method is applicable to order-agnostic cross-modal generation while achieving outstanding performance compared to the state-of-the-art multimodal methods. The source code for our method is available online https://anonymous.4open.science/r/SMVAE-9B3C/.

The dual-encoder has become the de facto architecture for dense retrieval. Typically, it computes the latent representations of the query and document independently, thus failing to fully capture the interactions between the query and document. To alleviate this, recent work expects to get query-informed representations of documents. During training, it expands the document with a real query, while replacing the real query with a generated pseudo query at inference. This discrepancy between training and inference makes the dense retrieval model pay more attention to the query information but ignore the document when computing the document representation. As a result, it even performs worse than the vanilla dense retrieval model, since its performance depends heavily on the relevance between the generated queries and the real query. In this paper, we propose a curriculum sampling strategy, which also resorts to the pseudo query at training and gradually increases the relevance of the generated query to the real query. In this way, the retrieval model can learn to extend its attention from the document only to both the document and query, hence getting high-quality query-informed document representations. Experimental results on several passage retrieval datasets show that our approach outperforms the previous dense retrieval methods1.

In this work, we study the black-box targeted attack problem from the model discrepancy perspective. On the theoretical side, we present a generalization error bound for black-box targeted attacks, which gives a rigorous theoretical analysis for guaranteeing the success of the attack. We reveal that the attack error on a target model mainly depends on empirical attack error on the substitute model and the maximum model discrepancy among substitute models. On the algorithmic side, we derive a new algorithm for black-box targeted attacks based on our theoretical analysis, in which we additionally minimize the maximum model discrepancy(M3D) of the substitute models when training the generator to generate adversarial examples. In this way, our model is capable of crafting highly transferable adversarial examples that are robust to the model variation, thus improving the success rate for attacking the black-box model. We conduct extensive experiments on the ImageNet dataset with different classification models, and our proposed approach outperforms existing state-of-the-art methods by a significant margin. Our codes will be released.

Contrastive deep graph clustering, which aims to divide nodes into disjoint groups via contrastive mechanisms, is a challenging research spot. Among the recent works, hard sample mining-based algorithms have achieved great attention for their promising performance. However, we find that the existing hard sample mining methods have two problems as follows. 1) In the hardness measurement, the important structural information is overlooked for similarity calculation, degrading the representativeness of the selected hard negative samples. 2) Previous works merely focus on the hard negative sample pairs while neglecting the hard positive sample pairs. Nevertheless, samples within the same cluster but with low similarity should also be carefully learned. To solve the problems, we propose a novel contrastive deep graph clustering method dubbed Hard Sample Aware Network (HSAN) by introducing a comprehensive similarity measure criterion and a general dynamic sample weighing strategy. Concretely, in our algorithm, the similarities between samples are calculated by considering both the attribute embeddings and the structure embeddings, better revealing sample relationships and assisting hardness measurement. Moreover, under the guidance of the carefully collected high-confidence clustering information, our proposed weight modulating function will first recognize the positive and negative samples and then dynamically up-weight the hard sample pairs while down-weighting the easy ones. In this way, our method can mine not only the hard negative samples but also the hard positive sample, thus improving the discriminative capability of the samples further. Extensive experiments and analyses demonstrate the superiority and effectiveness of our proposed method.