U-shaped networks are widely used in various medical image tasks, such as segmentation, restoration and reconstruction, but most of them usually rely on centralized learning and thus ignore privacy issues. To address the privacy concerns, federated learning (FL) and split learning (SL) have attracted increasing attention. However, it is hard for both FL and SL to balance the local computational cost, model privacy and parallel training simultaneously. To achieve this goal, in this paper, we propose Robust Split Federated Learning (RoS-FL) for U-shaped medical image networks, which is a novel hybrid learning paradigm of FL and SL. Previous works cannot preserve the data privacy, including the input, model parameters, label and output simultaneously. To effectively deal with all of them, we design a novel splitting method for U-shaped medical image networks, which splits the network into three parts hosted by different parties. Besides, the distributed learning methods usually suffer from a drift between local and global models caused by data heterogeneity. Based on this consideration, we propose a dynamic weight correction strategy (\textbf{DWCS}) to stabilize the training process and avoid model drift. Specifically, a weight correction loss is designed to quantify the drift between the models from two adjacent communication rounds. By minimizing this loss, a correction model is obtained. Then we treat the weighted sum of correction model and final round models as the result. The effectiveness of the proposed RoS-FL is supported by extensive experimental results on different tasks. Related codes will be released at https://github.com/Zi-YuanYang/RoS-FL.

Multi-source translation (MST), which typically receives multiple source sentences of the same meaning in different languages, has been shown superior to single-source translation. As the quantity of multi-source parallel data is limited, taking full advantage of single-source data and limited multi-source data to make models perform well when receiving as many as possible sources remains a challenge. Unlike previous work mostly devoted to supervised scenarios, we focus on zero-shot MST: expecting models to be able to process unseen combinations of multiple sources, e.g., unseen language combinations, during inference. We propose a simple yet effective parameter efficient method, named Prompt Gating, which appends prompts to the model inputs and attaches gates on the extended hidden states for each encoder layer. It shows strong zero-shot transferability (+9.0 BLEU points maximally) and remarkable compositionality (+15.6 BLEU points maximally) on MST, and also shows its superiorities over baselines on lexically constrained translation.

Although continually extending an existing NMT model to new domains or languages has attracted intensive interest in recent years, the equally valuable problem of continually improving a given NMT model in its domain by leveraging knowledge from an unlimited number of existing NMT models is not explored yet. To facilitate the study, we propose a formal definition for the problem named knowledge accumulation for NMT (KA-NMT) with corresponding datasets and evaluation metrics and develop a novel method for KA-NMT. We investigate a novel knowledge detection algorithm to identify beneficial knowledge from existing models at token level, and propose to learn from beneficial knowledge and learn against other knowledge simultaneously to improve learning efficiency. To alleviate catastrophic forgetting, we further propose to transfer knowledge from previous to current version of the given model. Extensive experiments show that our proposed method significantly and consistently outperforms representative baselines under homogeneous, heterogeneous, and malicious model settings for different language pairs.

With the evergrowing sizes of pre-trained models (PTMs), it has been an emerging practice to only provide the inference APIs for users, namely model-as-a-service (MaaS) setting. To adapt PTMs with model parameters frozen, most current approaches focus on the input side, seeking for powerful prompts to stimulate models for correct answers. However, we argue that input-side adaptation could be arduous due to the lack of gradient signals and they usually require thousands of API queries, resulting in high computation and time costs. In light of this, we present Decoder Tuning (DecT), which in contrast optimizes task-specific decoder networks on the output side. Specifically, DecT first extracts prompt-stimulated output scores for initial predictions. On top of that, we train an additional decoder network on the output representations to incorporate posterior data knowledge. By gradient-based optimization, DecT can be trained within several seconds and requires only one PTM query per sample. Empirically, we conduct extensive natural language understanding experiments and show that DecT significantly outperforms state-of-the-art algorithms with a $10^3\times$ speed-up.

提示将下游应用程序作为语言建模任务施放，与使用预训练的模型进行标准微调相比，已显示出样本有效的效率。但是，提示的一个陷阱是需要手动设计的模式，其结果可能是不直觉的，需要大量的验证集来调整。为了应对挑战，我们提出了一种全自动提示方法Autoseq：（1）我们在序列到序列模型上采用自然语言提示，从而实现自由形式生成和更大的标签搜索空间； （2）我们提出了标签序列 - 无限长度的短语以口头表达标签 - 这消除了手动模板的需求，并且比单个标签单词更具有表现力； （3）我们使用Beam Search自动生成大量的标签序列候选物，并提出对比度重新排列以获得最佳组合。 Autoseq显着胜过其他无手动设计方法，例如软提示调整，适配器调整和自动搜索单个标签单词；生成的标签序列比各种任务上的精选手动序列更好。我们的方法揭示了几次学习中序列模型的潜力，并阐明了通用通用和自动提示的途径。本文的源代码可以从https://github.com/thunlp/seq2seq-prompt获得。

在过去的几年中，在各种文本生成任务中见证了各种自动编码器的优势。但是，由于文本的顺序性质，自动回归解码器倾向于忽略潜在变量，然后降低到简单的语言模型，称为KL消失的问题，当VAE与基于变压器的结构结合时，这将进一步恶化。为了改善这个问题，我们提出了一种新型变化变压器框架Della。德拉（Della）从较低层的层中得知一系列层的潜在变量，每个变量都从下层的层中推断出，并通过低级张量产品与隐藏状态紧密耦合。通过这种方式，Della强迫这些后部潜在变量将其与整个计算路径深入融合，从而结合了更多信息。从理论上讲，我们可以将我们的方法视为纠缠潜在变量，以避免通过层减少后验信息，从而使DELLA即使没有任何退火或阈值技巧，也可以使DELLA获得更高的非零KL值。与多个强大的基线相比，对四个无条件和三个条件生成任务的实验表明，Della可以更好地减轻KL消失并改善质量和多样性。

深层自然语言处理（NLP）模型的快速发展导致迫切需要对这些模型单独提出的统一理解。由于缺乏解释低级（例如单词）和高级（例如，短语）特征的统一措施，现有方法无法满足一个框架中不同模型的需求。我们已经开发了一个视觉分析工具DeepNLPVI，以使对文本分类的NLP模型有统一的理解。关键思想是一种基于信息的度量，它提供了有关模型的每一层如何维护样本中输入单词信息的定量解释。我们在每个层的内部和界面信息中对单词对最终预测的重要性以及单词之间的关系（例如短语的形成）进行建模。多层可视化由语料库级，样本级别和单词级可视化组成，支持从整体训练集到单个样本的分析。关于分类任务和模型比较的两个案例研究表明，DeepNLPVI可以帮助用户有效地确定样本和模型架构引起的潜在问题，然后进行明智的改进。

文本后门攻击是对NLP系统的实际威胁。通过在训练阶段注入后门，对手可以通过预定义的触发器控制模型预测。由于已经提出了各种攻击和防御模型，因此进行严格的评估至关重要。但是，我们在以前的后门学习评估中重点介绍了两个问题：（1）忽略了现实世界情景（例如释放中毒的数据集或模型）之间的差异，我们认为每种情况都有其自身的限制和关注点，因此需要特定的评估。协议； （2）评估指标仅考虑攻击是否可以翻转模型对中毒样品的预测并保留对良性样品的表演，但是忽略了中毒样品也应该是隐秘和语义上的。为了解决这些问题，我们将现有作品分为三种实际情况，在这种情况下，攻击者分别释放数据集，预培训模型和微调模型，然后讨论其独特的评估方法。关于指标，为了完全评估中毒样本，我们使用语法误差增加和隐形性差异以及有效性的文本相似性。对框架进行正式化后，我们开发了一个开源工具包openbackdoor，以促进文本后门学习的实现和评估。使用此工具包，我们在建议的范式下进行基准攻击和防御模型进行广泛的实验。为了促进针对中毒数据集的不充分的防御能力，我们进一步提出了Cube，这是一个简单而强大的基于聚类的防御基线。我们希望我们的框架和基准可以作为未来模型开发和评估的基石。

通过微调调整大型预训练模型（PTM）会施加过刺激的计算和存储负担。对参数有效调整（PET）的最新研究发现，与常规微调相比，仅优化以PTM为条件的一小部分参数才能产生PAR性能。通常，PET方法精确设计参数有效的模块（PET模块）可以应用于PTMS内部的任意细粒位置。但是，这些细粒度位置的有效性很大程度上依赖于复杂的手动指定，因此通常会产生次优的结果。与手动指定相反，我们以自动方式探索构建宠物模块。我们将自动\ textbf {s} earch \ textbf {s} parse \ textbf {s} \ textbf {p} arameter- \ textbf {e} fficbf {e} fficient \ textbf {t textbf {t} uning（s $^3 $ pet） 。基于各种PET方法的统一框架，S $^3 $ PET通过双层优化进行了可区分的PET结构搜索，并提出了移动的全局Sigmoid方法，以明确控制可训练的参数的数量。广泛的实验表明，S $^3 $ PET超过了具有较低训练参数的手册和随机结构。搜索结构可保留99 \％的微调性能，具有0.01 \％可训练的参数。此外，S $^3 $ PET的优势通过极低的训练参数预算（0.0009 \％$ \ sim $ 0.01 \％）进行扩增。搜索结构是可转移和解释的，为PET方法的未来设计提供了建议和指导。

控制语言模型的主要方法在控制高级属性（例如主题和情感）方面具有突出性。但是，这些方法通常需要特定于条件的数据或计算昂贵。我们提出了一种新的简单引导解码方法，伽玛采样，该方法不需要任何培训数据来实现可控制的文本生成，同时保持快速生成速度。伽玛采样将与属性相关的信息（由人类或语言模型本身提供）引入采样过程中，以指导语言模型，以生成具有所需属性的文本。由于不涉及培训，因此可以轻松地将伽马抽样应用于任何语言模型以进行可控文本。通过实验，我们表明，伽马取样的GPT2-MALL（1.17亿）优于PPLM（345m）和CTRL（1.6B）的多样性，属性相关性以及生成样品的整体质量。