Narrative comprehension on long stories and novels has been a challenging domain attributed to their intricate plotlines and entangled, often evolving relations among characters and entities. Given the LLM's diminished reasoning over extended context and its high computational cost, retrieval-based approaches remain a pivotal role in practice. However, traditional RAG methods could fall short due to their stateless, single-step retrieval process, which often overlooks the dynamic nature of capturing interconnected relations within long-range context. In this work, we propose ComoRAG, holding the principle that narrative reasoning is not a one-shot process, but a dynamic, evolving interplay between new evidence acquisition and past knowledge consolidation, analogous to human cognition on reasoning with memory-related signals in the brain. Specifically, when encountering a reasoning impasse, ComoRAG undergoes iterative reasoning cycles while interacting with a dynamic memory workspace. In each cycle, it generates probing queries to devise new exploratory paths, then integrates the retrieved evidence of new aspects into a global memory pool, thereby supporting the emergence of a coherent context for the query resolution. Across four challenging long-context narrative benchmarks (200K+ tokens), ComoRAG outperforms strong RAG baselines with consistent relative gains up to 11% compared to the strongest baseline. Further analysis reveals that ComoRAG is particularly advantageous for complex queries requiring global comprehension, offering a principled, cognitively motivated paradigm for retrieval-based stateful reasoning.

Complex reasoning problems often involve implicit spatial and geometric relationships that are not explicitly encoded in text. While recent reasoning models perform well across many domains, purely text-based reasoning struggles to capture structural constraints in complex settings. In this paper, we introduce FIGR, which integrates executable visual construction into multi-turn reasoning via end-to-end reinforcement learning. Rather than relying solely on textual chains of thought, FIGR externalizes intermediate hypotheses by generating executable code that constructs diagrams within the reasoning loop. An adaptive reward mechanism selectively regulates when visual construction is invoked, enabling more consistent reasoning over latent global properties that are difficult to infer from text alone. Experiments on seven challenging mathematical benchmarks demonstrate that FIGR outperforms strong text-only chain-of-thought baselines, improving the base model by 13.12% on AIME 2025 and 11.00% on BeyondAIME. These results highlight the effectiveness of precise, controllable figure construction of FIGR in enhancing complex reasoning ability.

Zhiyu Xu, Lean Wang, Yuanxin Liu, Lei Li, Hao Zhou, Fandong Meng, Jie Zhou, Xu Sun. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Xue Zhang, Yunlong Liang, Fandong Meng, Songming Zhang, Kaiyu Huang, Yufeng Chen, Xu Jinan, Jie Zhou. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Abstract To enhance the efficiency of the attention mechanism within large language models (LLMs), previous works primarily compress the Key-Value cache or group attention heads, while largely overlooking redundancy between layers. Our comprehensive analyses across various LLMs show that highly similar attention patterns persist within most layers. It’s intuitive to reduce the redundancy by sharing attention weights across layers. However, further analysis reveals two challenges: (1) Directly sharing the weight matrix without carefully rearranging the attention heads proves to be ineffective; (2) Shallow layers are vulnerable to small deviations in attention weights. Driven by these insights, we introduce LiSA, a lightweight substitute for self-attention in well-trained LLMs. LiSA employs tiny feed-forward networks to align attention heads between adjacent layers and low-rank matrices to approximate differences in layer-wise attention weights. Evaluations encompassing 13 typical benchmarks demonstrate that LiSA maintains high response quality in terms of accuracy and perplexity while reducing redundant attention calculations within 53% −84% of the total layers. Our implementations of LiSA achieve a 6 × compression of Q and K matrices within the attention mechanism, with maximum throughput improvements 19.5%, 32.3%, and 40.1% for LLaMA3-8B, LLaMA2-7B, and LLaMA2-13B, respectively. Our code is available at https://github.com/takagi97/lisa.

Abstract Recently, deep reasoning LLMs (e.g., OpenAI o1 and DeepSeek-R1) have shown promising performance in various downstream tasks. Free translation is an important and interesting task in the multilingual world, which requires going beyond word-for-word translation. However, the task is still under-explored in deep reasoning LLMs. In this paper, we introduce DeepTrans, a deep reasoning translation model that learns free translation via reinforcement learning (RL). Specifically, we carefully build a reward model with pre-defined scoring criteria on both the translation results and the thought processes. The reward model teaches DeepTrans how to think and free-translate the given sentences during RL. Besides, our RL training does not need any labeled translations, avoiding the human-intensive annotation or resource-intensive data synthesis. Experimental results show the effectiveness of DeepTrans. Using Qwen2.5-7B as the backbone, DeepTrans improves performance by 16.3% in literature translation, and outperforms strong deep reasoning LLMs. Moreover, we summarize the failures and interesting findings during our RL exploration. We hope this work could inspire other researchers in free translation.1

Recommendations usually focus on immediate accuracy metrics like Click-Through Rate (CTR), ignoring user long-term metrics. User retention, which reflects the percentage of today’s users who will return to the system in the next few days, should be paid more attention to. However, most existing methods did not focus on user retention, since their complexity and uncertainty make it extremely hard to discover why a user will or will not return to a system. Recently, a few pioneers have optimized user retention, focusing solely on accuracy without delving into its underlying rationale. This is primarily due to the absence of explicit supervised signals. In this work, we design a Behavior-wise Contrastive Multi-Instance Learning (BCMIL) module, which jointly models clicked and impressed items to capture interpretable user retention. Specifically, we conduct in-depth analyses in real-world scenarios to discover implicit retention-related supervised signals. To model these signals, we design a Forward Supervised Signals Extractor (FSSE) that utilizes a heterogeneous graph, enhancing the reliability of user retention. To mitigate randomness and uncertainty, we propose a Backward Supervised Signals Stabilizer (BSSS) that utilizes overlooked label-part behaviors within each training window to retrospectively guide the training process. Offline and online evaluations of an industrial system verify the effectiveness of our methods.

While long-context inference is crucial for advancing large language model (LLM) applications, its prefill speed remains a significant bottleneck. Current approaches, including sequence parallelism strategies and compute reduction through approximate attention mechanisms, still fall short of delivering optimal inference efficiency. This hinders scaling the inputs to longer sequences and processing long-context queries in a timely manner. To address this, we introduce APB, an efficient long-context inference framework that leverages multi-host approximate attention to enhance prefill speed by reducing compute and enhancing parallelism simultaneously. APB introduces a communication mechanism for essential key-value pairs within a sequence parallelism framework, enabling a faster inference speed while maintaining task performance. We implement APB by incorporating a tailored FlashAttn kernel alongside optimized distribution strategies, supporting diverse models and parallelism configurations. APB achieves speedups of up to 9.2\times, 4.2\times, and 1.6\times compared with FlashAttn, RingAttn, and StarAttn, respectively, without any observable task performance degradation.

The sparse Mixture-of-Experts (MoE) has achieved significant progress for neural machine translation (NMT). However, there exist two limitations in current MoE solutions which may lead to sub-optimal performance: 1) they directly use the task knowledge of NMT into MoE (e.g., domain/linguistics-specific knowledge), which are generally unavailable at practical application and neglect the naturally grouped domain/linguistic properties; 2) the expert selection only depends on the localized token representation without considering the context, which fully grasps the state of each token in a global view. To address the above limitations, we propose THOR-MoE via arming the MoE with hierarchical task-guided and context-responsive routing policies. Specifically, it 1) firstly predicts the domain/language label and then extracts mixed domain/language representation to allocate task-level experts in a hierarchical manner; 2) injects the context information to enhance the token routing from the pre-selected task-level experts set, which can help each token to be accurately routed to more specialized and suitable experts. Extensive experiments on multi-domain translation and multilingual translation benchmarks with different architectures consistently demonstrate the superior performance of THOR-MoE. Additionally, the THOR-MoE operates as a plug-and-play module compatible with existing Top-(CITATION) or Top-(CITATION) routing schemes, ensuring broad applicability across diverse MoE architectures. For instance, compared with vanilla Top- (CITATION) routing, the context-aware manner can achieve an average improvement of 0.75 BLEU with less than 22% activated parameters on multi-domain translation tasks.

Kun Ouyang, Yuanxin Liu, Shicheng Li, Yi Liu, Hao Zhou, Fandong Meng, Jie Zhou, Xu Sun. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Jiaan Wang, Fandong Meng, Zengkui Sun, Yunlong Liang, Yuxuan Cao, Jiarong Xu, Haoxiang Shi, Jie Zhou. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Continually expanding new languages for existing large language models (LLMs) is a promising yet challenging approach to building powerful multilingual LLMs.The biggest challenge is to make the model continuously learn new languages while preserving the proficient ability of old languages.To achieve this, recent work utilizes the Mixture-of-Experts (MoE) architecture to expand new languages by adding new experts and avoid catastrophic forgetting of old languages by routing corresponding tokens to the original model backbone (old experts).Although intuitive, this kind of method is parameter-costly when expanding new languages and still inevitably impacts the performance of old languages.To address these limitations, we analyze the language characteristics of different layers in LLMs and propose a layer-wise expert allocation algorithm (LayerMoE) to determine the appropriate number of new experts for each layer.Specifically, we find different layers in LLMs exhibit different representation similarities between languages and then utilize the similarity as the indicator to allocate experts for each layer, i.e., the higher similarity, the fewer experts.Additionally, to further mitigate the forgetting of old languages, we add a classifier in front of the router network on the layers with higher similarity to guide the routing of old language tokens.Experimental results show that our method outperforms the previous state-of-the-art baseline with 60% fewer experts in the single-expansion setting and with 33.3% fewer experts in the lifelong-expansion setting, demonstrating the effectiveness of our method.

Recent studies have explored Continual Instruction Tuning (CIT) in Multimodal Large Language Models (MLLMs), with a primary focus on Task-incremental CIT, where MLLMs are required to continuously acquire new tasks.However, the more practical and challenging Domain-incremental CIT, focused on the continual adaptation of MLLMs to new domains, remains underexplored.In this paper, we propose a new Sparse Mixture of Expert (SMoE) based method for domain-incremental CIT in MLLMs.During training, we learn a domainspecific SMoE module for each new domain in every FFN sub-layer of MLLMs, preventing catastrophic forgetting caused by inter-domain conflicts.Moreover, we equip the SMoE module with a domain-specific autoregressive loss (DSAL), which is used to identify the most suitable SMoE module for processing each test instruction during inference.To further enhance the SMoE module's ability to learn domain knowledge, we design an adaptive thresholdbased router (AT-Router) that allocates computing resources (experts) to instruction tokens based on their importance.Finally, we establish a new benchmark to evaluate the efficacy of our method and advance future research.Extensive experiments show that our method consistently outperforms all competitive baselines.

The few-shot relation extraction (FSRE) aims at enhancing the model's generalization to new relations with a few labeled instances.Existing studies usually adopt prototype networks (Pro-toNets) for FSRE and assume that the support set, adapting the model to new relations, only contains accurately labeled support instances.However, this assumption is often unrealistic, as even carefully-annotated datasets commonly contain mislabeled instances.In this paper, we first conduct a preliminary study that reveals the high sensitivity of ProtoNets to noisy labels in the support set.Moreover, we find that fully leveraging mislabeled support instances is crucial for enhancing the FSRE model robustness.Thus, we propose a self-denoising model for FSRE, designed to improve model robustness by automatically correcting mislabeled support instances.It comprises two core components: 1) a label correction module (LCM), used to correct noisy labels of support instances based on the distances between them in the embedding space, and 2) a relation classification module (RCM), aimed to achieve more accurate predictions for new relations based on the corrected labels produced by LCM.Moreover, we propose a feedback-based training strategy, which focuses on training LCM and RCM to synergistically handle noisy labels in support set.Experimental results on two public datasets confirm the robustness of our model.Particularly, even in scenarios without noisy labels, our model significantly outperforms all baselines.

Weilin Zhao, Tengyu Pan, Xu Han, Yudi Zhang, Sun Ao, Yuxiang Huang, Kaihuo Zhang, Weilun Zhao, Yuxuan Li, Jie Zhou, Hao Zhou, Jianyong Wang, Maosong Sun, Zhiyuan Liu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Theory-of-Mind (ToM) is a fundamental psychological capability that allows humans to understand and interpret the mental states of others. Humans infer others’ thoughts by integrating causal cues and indirect clues from broad contextual information, often derived from past interactions. In other words, human ToM heavily relies on the understanding about the backgrounds and life stories of others. Unfortunately, this aspect is largely overlooked in existing benchmarks for evaluating machines’ ToM capabilities, due to their usage of short narratives without global context, especially personal background of characters. In this paper, we verify the importance of comprehensive contextual understanding about personal backgrounds in ToM and assess the performance of LLMs in such complex scenarios. To achieve this, we introduce CharToM-QA benchmark, comprising 1,035 ToM questions based on characters from classic novels. Our human study reveals a significant disparity in performance: the same group of educated participants performs dramatically better when they have read the novels compared to when they have not. In parallel, our experiments on state-of-the-art LLMs, including the very recent o1 and DeepSeek-R1 models, show that LLMs still perform notably worse than humans, despite that they have seen these stories during pre-training. This highlights the limitations of current LLMs in capturing the nuanced contextual information required for ToM reasoning.

Large language models (LLMs) have shown incredible performance in completing various real-world tasks. The current paradigm of knowledge learning for LLMs is mainly based on learning from examples, in which LLMs learn the internal rule implicitly from a certain number of supervised examples. However, this learning paradigm may not well learn those complicated rules, especially when the training examples are limited. We are inspired that humans can learn the new tasks or knowledge in another way by learning from rules. That is, humans can learn new tasks or grasp new knowledge quickly and generalize well given only a detailed rule and a few optional examples. Therefore, in this paper, we aim to explore the feasibility of this new learning paradigm, which targets on encoding rule-based knowledge into LLMs. We further propose rule distillation, which first uses the strong in-context abilities of LLMs to extract the knowledge from the textual rules, and then explicitly encode the knowledge into the parameters of LLMs by learning from the above in-context signals produced inside the model. Our experiments show that making LLMs learn from rules by our method is much more efficient than example-based learning in both the sample size and generalization ability. Warning: This paper may contain examples with offensive content.

Multilingual knowledge editing (MKE) aims to simultaneously update factual knowledge across multiple languages within large language models (LLMs). Previous research indicates that the same knowledge across different languages within LLMs exhibits a degree of shareability. However, most existing MKE methods overlook the connections of the same knowledge between different languages, resulting in knowledge conflicts and limited edit performance. To address this issue, we first investigate how LLMs process multilingual factual knowledge and discover that the same factual knowledge in different languages generally activates a shared set of neurons, which we call language-agnostic factual neurons (LAFNs). These neurons represent the same factual knowledge shared across languages and imply the semantic connections among multilingual knowledge. Inspired by this finding, we propose a new MKE method by Locating and Updating Language-Agnostic Factual Neurons (LU-LAFNs) to edit multilingual knowledge simultaneously, which avoids knowledge conflicts and thus improves edit performance. Experimental results on Bi-ZsRE and MzsRE benchmarks demonstrate that our method achieves the best edit performance, indicating the effectiveness and importance of modeling the semantic connections among multilingual knowledge.

The advancement of Generative Adversarial Networks (GANs) and diffusion models significantly enhances the realism of synthetic images, driving progress in image processing and creative design. However, this progress also necessitates the development of effective detection methods, as synthetic images become increasingly difficult to distinguish from real ones. This difficulty leads to societal issues, such as the spread of misinformation, identity theft, and online fraud. While previous detection methods perform well on public benchmarks, they struggle with our benchmark, FakeART, particularly when dealing with the latest models and cross-domain tasks (e.g., photo-to-painting). To address this challenge, we develop a new synthetic image detection technique based on color distribution. Unlike real images, synthetic images often exhibit uneven color distribution. By employing color quantization and restoration techniques, we analyze the color differences before and after image restoration. We discover and prove that these differences closely relate to the uniformity of color distribution. Based on this finding, we extract effective color features and combine them with image features to create a detection model with only 1.4 million parameters. This model achieves state-of-the-art results across various evaluation benchmarks, including the challenging FakeART dataset.

Yuan Liu, Zhongyin Zhao, Le Tian, Haicheng Wang, Xubing Ye, Yangxiu You, Zilin Yu, Chuhan Wu, Zhou Xiao, Yang Yu, Jie Zhou. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Ziqing Qiao, Yongheng Deng, Jiali Zeng, Dong Wang, Lai Wei, Guanbo Wang, Fandong Meng, Jie Zhou, Ju Ren, Yaoxue Zhang. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Structural information in images is crucial for aesthetic assessment, and it is widely recognized in the artistic field that imitating the structure of other works significantly infringes on creators’ rights. The advancement of diffusion models has led to AI-generated content imitating artists’ structural creations, yet effective detection methods are still lacking. In this paper, we define this phenomenon as "structural infringement" and propose a corresponding detection method. Additionally, we develop quantitative metrics and create manually annotated datasets for evaluation: the SIA dataset of synthesized data, and the SIR dataset of real data. Due to the current lack of datasets for structural infringement detection, we propose a new data synthesis strategy based on diffusion models and LLM, successfully training a structural infringement detection model. Experimental results show that our method can successfully detect structural infringements and achieve notable improvements on annotated test sets.

High-quality animated stickers usually contain transparent channels, which are often ignored by current video generation models. To generate fine-grained animated transparency channels, existing methods can be roughly divided into video matting algorithms and diffusion-based algorithms. The methods based on video matting have poor performance in dealing with semi-open areas in stickers, while diffusion-based methods are often used to model a single image, which will lead to local flicker when modeling animated stickers. In this paper, we firstly propose an ILDiff method to generate animated transparent channels through implicit layout distillation, which solves the problems of semi-open area collapse and no consideration of temporal information in existing methods. Secondly, we create the Transparent Animated Sticker Dataset (TASD), which contains 0.32M high-quality samples with transparent channel, to provide data support for related fields. Extensive experiments demonstrate that ILDiff can produce finer and smoother transparent channels compared to other methods such as Matting Anything and Layer Diffusion. Our code and dataset will be released at link https://xiaoyuan1996.github.io.

Vision-language models (VLMs) often struggle with compositional reasoning due to insufficient high-quality image-text data. To tackle this challenge, we propose a novel block-based diffusion approach that automatically generates counterfactual datasets without manual annotation. Our method utilizes large language models to identify entities and their spatial relationships. It then independently generates image blocks as "puzzle pieces" coherently arranged according to specified compositional rules. This process creates diverse, high-fidelity counterfactual image-text pairs with precisely controlled variations. In addition, we introduce a specialized loss function that differentiates inter-set from intra-set samples, enhancing training efficiency and reducing the need for negative samples. Experiments demonstrate that fine-tuning VLMs with our counterfactual datasets significantly improves visual reasoning performance. Our approach achieves state-of-the-art results across multiple benchmarks while using substantially less training data than existing methods.

Approximately 200 million individuals around the world suffer from varying degrees of visual impairment, making it crucial to leverage AI technology to offer walking assistance for these people. With the recent progress of vision-language models (VLMs), employing VLMs to improve this field has emerged as a popular research topic. However, most existing methods are studied on self-built question-answering datasets, lacking a unified training and testing benchmark for walk guidance. Moreover, in blind walking task, it is necessary to perform real-time streaming video parsing and generate concise yet informative reminders, which poses a great challenge for VLMs that suffer from redundant responses and low inference efficiency. In this paper, we firstly release a diverse, extensive, and unbiased walking awareness dataset, containing 12k video-manual annotation pairs from Europe and Asia to provide a fair training and testing benchmark for blind walking task. Furthermore, a WalkVLM model is proposed, which employs chain of thought for hierarchical planning to generate concise but informative reminders and utilizes temporal-aware adaptive prediction to reduce the temporal redundancy of reminders. Finally, we have established a solid benchmark for blind walking task and verified the advantages of WalkVLM in stream video processing for this task compared to other VLMs. Our dataset and code will be released at anonymous link https://walkvlm2024.github.io.

Knowledge distillation (KD) is widely used to train small, high-performing student language models (LMs) using large teacher LMs. While effective in fine-tuning, KD during pre-training faces efficiency, flexibility, and effectiveness issues. Existing methods either incur high computational costs due to online teacher inference, require tokenization matching between teacher and student LMs, or risk losing the difficulty and diversity of the teacher-generated training data. In this work, we propose **MiniPLM**, a KD framework for pre-training LMs by refining the training data distribution with the teacher LM's knowledge. For efficiency, MiniPLM performs offline teacher inference, allowing KD for multiple student LMs without adding training costs. For flexibility, MiniPLM operates solely on the training corpus, enabling KD across model families. For effectiveness, MiniPLM leverages the differences between large and small LMs to enhance the training data difficulty and diversity, helping student LMs acquire versatile and sophisticated knowledge. Extensive experiments demonstrate that MiniPLM boosts the student LMs' performance on 9 common downstream tasks, improves language modeling capabilities, and reduces pre-training computation. The benefit of MiniPLM extends to larger training scales, evidenced by the scaling curve extrapolation. Further analysis reveals that MiniPLM supports KD across model families and enhances the pre-training data utilization. Our code, data, and models can be found at https://github.com/thu-coai/MiniPLM.

The prohibitive training costs of Large Language Models (LLMs) have emerged as a significant bottleneck in the development of next-generation LLMs. In this paper, we show that it is possible to significantly reduce the training costs of LLMs without sacrificing their performance. Specifically, we introduce patch-level training for LLMs, in which multiple tokens are aggregated into a unit of higher information density, referred to as a `patch', to serve as the fundamental text unit for training LLMs. During patch-level training, we feed the language model shorter sequences of patches and train it to predict the next patch, thereby processing the majority of the training data at a significantly reduced cost. Following this, the model continues token-level training on the remaining training data to align with the inference mode. Experiments on a diverse range of models (370M-2.7B parameters) demonstrate that patch-level training can reduce the overall training costs to 0.5$\times$, without compromising the model performance compared to token-level training. Source code: \url{https://github.com/shaochenze/PatchTrain}.

Large language models (LLMs) have achieved reasonable quality improvements in machine translation (MT). However, most current research on MT-LLMs still faces significant challenges in maintaining translation consistency and accuracy when processing entire documents. In this paper, we introduce DelTA, a Document-levEL Translation Agent designed to overcome these limitations. DelTA features a multi-level memory structure that stores information across various granularities and spans, including Proper Noun Records, Bilingual Summary, Long-Term Memory, and Short-Term Memory, which are continuously retrieved and updated by auxiliary LLM-based components. Experimental results indicate that DelTA significantly outperforms strong baselines in terms of translation consistency and quality across four open/closed-source LLMs and two representative document translation datasets, achieving an increase in consistency scores by up to 4.58 percentage points and in COMET scores by up to 3.16 points on average. DelTA employs a sentence-by-sentence translation strategy, ensuring no sentence omissions and offering a memory-efficient solution compared to the mainstream method. Furthermore, DelTA improves pronoun and context-dependent translation accuracy, and the summary component of the agent also shows promise as a tool for query-based summarization tasks. The code and data of our approach are released at https://github.com/YutongWang1216/DocMTAgent.

media, Nowadays, advanced text-to-video algorithms have spawned numerous general video generation systems that allow users to customize high-quality, photo-realistic videos by only providing simple text prompts. However, creating customized animated stickers, which have lower frame rates and more abstract semantics than videos, is greatly hindered by difficulties in data acquisition and incomplete benchmarks. To facilitate the exploration of researchers in animated sticker generation (ASG) field, we construct the currently largest vision-language sticker dataset named "VSD2M" at a two-million scale that contains static and animated stickers. Furthermore, to improve the performance of traditional video generation methods on ASG tasks with discrete characteristics, we propose a Spatial Temporal Interaction layer that utilizes semantic interaction and detail preservation to address the issue of insufficient information utilization. To our knowledge, this is the most comprehensive benchmark for multi-frame ASG task, and we hope it can provide valuable inspiration for other scholars in intelligent creation.

Due to distribution shift, the performance of deep learning-based method for image dehazing is adversely affected when applied to real-world hazy images. In this paper, we find that such deviation in dehazing task between real and synthetic domains may come from the imperfect collection of clean data. Owing to the complexity of the scene and the effect of depth, the collected clean data cannot strictly meet the ideal conditions, which makes the atmospheric physics model in the real domain inconsistent with that in the synthetic domain. For this reason, we come up with a synthetic-to-real dehazing method based on domain unification, which attempts to unify the relationship between the real and synthetic domain, thus to let the dehazing model more in line with the actual situation. Extensive experiments qualitatively and quantitatively demonstrate that the proposed dehazing method significantly outperforms state-of-the-art methods on real-world images.

Conventional wisdom suggests that autoregressive models are used to process discrete data. When applied to continuous modalities such as visual data, Visual AutoRegressive modeling (VAR) typically resorts to quantization-based approaches to cast the data into a discrete space, which can introduce significant information loss. To tackle this issue, we introduce a Continuous VAR framework that enables direct visual autoregressive generation without vector quantization. The underlying theoretical foundation is strictly proper scoring rules, which provide powerful statistical tools capable of evaluating how well a generative model approximates the true distribution. Within this framework, all we need is to select a strictly proper score and set it as the training objective to optimize. We primarily explore a class of training objectives based on the energy score, which is likelihood-free and thus overcomes the difficulty of making probabilistic predictions in the continuous space. Previous efforts on continuous autoregressive generation, such as GIVT and diffusion loss, can also be derived from our framework using other strictly proper scores. Source code: \url{https://github.com/shaochenze/EAR}.

Medical education relies heavily on Simulated Patients (SPs) to provide a safe environment for students to practice clinical skills, including medical image analysis. However, the high cost of recruiting qualified SPs and the lack of diverse medical imaging datasets have presented significant challenges. To address these issues, this paper introduces MedDiT, a novel knowledge-controlled conversational framework that can dynamically generate plausible medical images aligned with simulated patient symptoms, enabling diverse diagnostic skill training. Specifically, MedDiT integrates various patient Knowledge Graphs (KGs), which describe the attributes and symptoms of patients, to dynamically prompt Large Language Models' (LLMs) behavior and control the patient characteristics, mitigating hallucination during medical conversation. Additionally, a well-tuned Diffusion Transformer (DiT) model is incorporated to generate medical images according to the specified patient attributes in the KG. In this paper, we present the capabilities of MedDiT through a practical demonstration, showcasing its ability to act in diverse simulated patient cases and generate the corresponding medical images. This can provide an abundant and interactive learning experience for students, advancing medical education by offering an immersive simulation platform for future healthcare professionals. The work sheds light on the feasibility of incorporating advanced technologies like LLM, KG, and DiT in education applications, highlighting their potential to address the challenges faced in simulated patient-based medical education.

In many practical natural language applications, user data are highly sensitive, requiring anonymous uploads of text data from mobile devices to the cloud without user identifiers. However, the absence of user identifiers restricts the ability of cloud-based language models to provide personalized services, which are essential for catering to diverse user needs. The trivial method of replacing an explicit user identifier with a static user embedding as model input still compromises data anonymization. In this work, we propose to let each mobile device maintain a user-specific distribution to dynamically generate user embeddings, thereby breaking the one-to-one mapping between an embedding and a specific user. We further theoretically demonstrate that to prevent the cloud from tracking users via uploaded embeddings, the local distributions of different users should either be derived from a linearly dependent space to avoid identifiability or be close to each other to prevent accurate attribution. Evaluation on both public and industrial datasets using different language models reveals a remarkable improvement in accuracy from incorporating anonymous user embeddings, while preserving real-time inference requirement.

Recent advances in generative modeling have enabled the synthesis of high-quality artistic images. Nevertheless, systematic evaluation of generative models from an aesthetic standpoint is still lacking, which hinders progress in artistic image synthesis. Existing evaluation metrics, such as Fréchet Inception Distance (FID) and CMMD, struggle with aesthetic assessment: they rely on pretrained visual features that overlook nuanced artistic attributes and employ distance functions ill-suited for modeling the diverse, multi-modal distribution of artistic styles. To address these limitations, we propose ArtFRD, a metric specifically designed for generative aesthetic evaluation. Grounded in aesthetic theory, ArtFRD extracts visual features along four key aesthetic dimensions-brushstroke, composition, lighting, and color-to capture fine-grained artistic properties. To model the multi-modal nature of artistic styles, we adopt a Gaussian Mixture Model assumption and derive an efficient approximation of the Fisher-Rao distance, which serves as the final evaluation score. Extensive experiments demonstrate that ArtFRD aligns significantly better with human aesthetic judgments than existing metrics, even across a wide range of artistic styles. These results highlight its potential as a robust and interpretable foundation for future research in generative aesthetic evaluation.

Junjie Wu, Mo Yu, Lemao Liu, Dit-Yan Yeung, Jie Zhou. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Mo Yu, Lemao Liu, Junjie Wu, Tsz Ting Chung, Shunchi Zhang, Jiangnan Li, Dit-Yan Yeung, Jie Zhou. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

We introduce \emph{SLED}, an alternative approach to speech language modeling by encoding speech waveforms into sequences of continuous latent representations and modeling them autoregressively using an energy distance objective. The energy distance offers an analytical measure of the distributional gap by contrasting simulated and target samples, enabling efficient training to capture the underlying continuous autoregressive distribution. By bypassing reliance on residual vector quantization, SLED avoids discretization errors and eliminates the need for the complicated hierarchical architectures common in existing speech language models. It simplifies the overall modeling pipeline while preserving the richness of speech information and maintaining inference efficiency. Empirical results demonstrate that SLED achieves strong performance in both zero-shot and streaming speech synthesis, showing its potential for broader applications in general-purpose speech language models. Demos and code are available at \url{https://github.com/ictnlp/SLED-TTS}.

Multimodal language analysis is a rapidly evolving field that leverages multiple modalities to enhance the understanding of high-level semantics underlying human conversational utterances. Despite its significance, little research has investigated the capability of multimodal large language models (MLLMs) to comprehend cognitive-level semantics. In this paper, we introduce MMLA, a comprehensive benchmark specifically designed to address this gap. MMLA comprises over 61K multimodal utterances drawn from both staged and real-world scenarios, covering six core dimensions of multimodal semantics: intent, emotion, dialogue act, sentiment, speaking style, and communication behavior. We evaluate eight mainstream branches of LLMs and MLLMs using three methods: zero-shot inference, supervised fine-tuning, and instruction tuning. Extensive experiments reveal that even fine-tuned models achieve only about 60~70% accuracy, underscoring the limitations of current MLLMs in understanding complex human language. We believe that MMLA will serve as a solid foundation for exploring the potential of large language models in multimodal language analysis and provide valuable resources to advance this field. The datasets and code are open-sourced at https://github.com/thuiar/MMLA.

With the thriving of the pre-trained language model (PLM) widely verified in various NLP tasks, pioneer efforts attempt to explore the possible cooperation of the general textual information in PLM with the personalized behavioral information in user historical behavior sequences to enhance sequential recommendation (SR). However, despite the commonalities of input format and task goal, there are huge gaps between the behavioral and textual information, which obstruct thoroughly modeling SR as language modeling via PLM. To bridge the gap, we propose a novel unified pre-trained language model enhanced sequential recommendation (UPSR) that thoroughly transfers the next item prediction task to a text generation task, aiming to build a unified pre-trained recommendation model for multi-domain recommendation tasks. We formally design five key indicators, namely naturalness, domain consistency, informativeness, noise and ambiguity, and text length, to guide the text \(\rightarrow\) item adaptation (selecting appropriate text to form the item textual representation) and behavior sequence \(\rightarrow\) text sequence adaptation (transferring the sequence of item textual representations into a text sequence) differently for pre-training and fine-tuning stages, which are essential but under-explored by previous works. In experiments, we conduct extensive evaluations on seven datasets with both supervised and zero-shot settings and achieve the overall best performance. Comprehensive model analyses also provide valuable insights for behavior modeling via PLM, shedding light on large pre-trained recommendation models. The source codes will be released in the future.

Knowledge retrieval with multi-modal queries plays a crucial role in supporting knowledge-intensive multi-modal applications. However, existing methods face challenges in terms of their effectiveness and training efficiency, especially when it comes to training and integrating multiple retrievers to handle multi-modal queries. In this paper, we propose an innovative end-to-end generative framework for multi-modal knowledge retrieval. Our framework takes advantage of the fact that large language models (LLMs) can effectively serve as virtual knowledge bases, even when trained with limited data. We retrieve knowledge via a two-step process: 1) generating knowledge clues related to the queries, and 2) obtaining the relevant document by searching databases using the knowledge clue. In particular, we first introduce an object-aware prefix-tuning technique to guide multi-grained visual learning. Then, we align multi-grained visual features into the textual feature space of the LLM, employing the LLM to capture cross-modal interactions. Subsequently, we construct instruction data with a unified format for model training. Finally, we propose the knowledge-guided generation strategy to impose prior constraints in the decoding steps, thereby promoting the generation of distinctive knowledge clues. Through experiments conducted on three benchmarks, we demonstrate significant improvements ranging from 3.0% to 14.6% across all evaluation metrics when compared to strong baselines.

Open-sourced large language models (LLMs) have demonstrated remarkable efficacy in various tasks with instruction tuning. However, these models can sometimes struggle with tasks that require more specialized knowledge such as translation. One possible reason for such deficiency is that instruction tuning aims to generate fluent and coherent text that continues from a given instruction without being constrained by any task-specific requirements. Moreover, it can be more challenging to tune smaller LLMs with lower-quality training data. To address this issue, we propose a novel framework using examples in comparison to teach LLMs to learn translation. Our approach involves output comparison and preference comparison, presenting the model with carefully designed examples of correct and incorrect translations and an additional preference loss for better regularization. Empirical evaluation on four language directions of WMT2022 and FLORES-200 benchmarks shows the superiority of our proposed method over existing methods. Our findings offer a new perspective on fine-tuning LLMs for translation tasks and provide a promising solution for generating high-quality translations. Please refer to Github for more details: https://github.com/lemon0830/TIM.

Large language models (LLMs) have recently demonstrated remarkable performance across various Natual Language Processing tasks. In the field of multi-hop reasoning, the Chain-of-thought (CoT) prompt method has emerged as a paradigm, using curated stepwise reasoning demonstrations to enhance LLM's ability to reason and produce coherent rational pathways. To ensure the accuracy, reliability, and traceability of the generated answers, many studies have incorporated information retrieval (IR) to provide LLMs with external knowledge. However, existing CoT with IR methods decomposes questions into sub-questions based on a single compositionality type, which limits their effectiveness for questions involving multiple compositionality types. Additionally, these methods suffer from inefficient retrieval, as complex questions often contain abundant information, leading to the retrieval of irrelevant information inconsistent with the query's intent. In this work, we propose a novel question decomposition framework called TRQA for multi-hop question answering, which addresses these limitations. Our framework introduces a reasoning tree (RT) to represent the structure of complex questions. It consists of four components: the Reasoning Tree Constructor (RTC), the Question Generator (QG), the Retrieval and LLM Interaction Module (RAIL), and the Answer Aggregation Module (AAM). Specifically, the RTC predicts diverse sub-question structures to construct the reasoning tree, allowing a more comprehensive representation of complex questions. The QG generates sub-questions for leaf-node in the reasoning tree, and we explore two methods for QG: prompt-based and T5-based approaches. The IR module retrieves documents aligned with sub-questions, while the LLM formulates answers based on the retrieved information. Finally, the AAM aggregates answers along the reason tree, producing a definitive response from bottom to top.

In this paper, we present CSCD-NS, the first Chinese spelling check (CSC) dataset designed for native speakers, containing 40,000 samples from a Chinese social platform.Compared with existing CSC datasets aimed at Chinese learners, CSCD-NS is ten times larger in scale and exhibits a distinct error distribution, with a significantly higher proportion of word-level errors.To further enhance the data resource, we propose a novel method that simulates the input process through an input method, generating large-scale and high-quality pseudo data that closely resembles the actual error distribution and outperforms existing methods.Moreover, we investigate the performance of various models in this scenario, including large language models (LLMs), such as ChatGPT.The result indicates that generative models underperform BERT-like classification models due to strict length and pronunciation constraints.The high prevalence of word-level errors also makes CSC for native speakers challenging enough, leaving substantial room for improvement.

Incrementally expanding the capability of an existing translation model to solve new domain tasks over time is a fundamental and practical problem, which usually suffers from catastrophic forgetting.Generally, multi-domain learning can be seen as a good solution.However, there are two drawbacks: 1) it requires having the training data for all domains available at the same time, which may be unrealistic due to storage or privacy concerns; 2) it requires re-training the model on the data of all domains from scratch when adding a new domain and this is time-consuming and computationally expensive.To address these issues, we present a semi-supervised contrastive distillation framework for incremental neural machine translation.Specifically, to avoid catastrophic forgetting, we propose to exploit unlabeled data from the same distributions of the older domains through knowledge distillation.Further, to ensure the distinct domain characteristics in the model as the number of domains increases, we devise a cross-domain contrastive objective to enhance the distilled knowledge.Extensive experiments on domain translation benchmarks show that our approach, without accessing any previous training data or re-training on all domains from scratch, can significantly prevent the model from forgetting previously learned knowledge while obtaining good performance on the incrementally added domains.

Cheng Qian, Bingxiang He, Zhong Zhuang, Jia Deng, Yujia Qin, Xin Cong, Zhong Zhang, Jie Zhou, Yankai Lin, Zhiyuan Liu, Maosong Sun. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Neural Machine Translation (NMT) has made remarkable progress over the past years.However, under-translation and over-translation remain two challenging problems in state-of-theart NMT systems.In this work, we conduct an in-depth analysis on the underlying cause of under-translation in NMT, providing an explanation from the perspective of decoding objective.To optimize the beam search objective, the model tends to overlook words it is less confident about, leading to the under-translation phenomenon.Correspondingly, the model's confidence in predicting the End Of Sentence (EOS) diminishes when under-translation occurs, serving as a mild penalty for under-translated candidates.Building upon this analysis, we propose employing the confidence of predicting EOS as a detector for under-translation, and strengthening the confidence-based penalty to penalize candidates with a high risk of under-translation.Experiments on both synthetic and real-world data show that our method can accurately detect and rectify under-translated outputs, with minor impact on other correct translations.

Xiaozhi Wang, Hao Peng, Yong Guan, Kaisheng Zeng, Jianhui Chen, Lei Hou, Xu Han, Yankai Lin, Zhiyuan Liu, Ruobing Xie, Jie Zhou, Juanzi Li. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Large language models (LLMs) have exhibited remarkable performance in various natural language processing tasks.Techniques like instruction tuning have effectively enhanced the proficiency of LLMs in the downstream task of machine translation.However, the existing approaches fail to yield satisfactory translation outputs that match the quality of supervised neural machine translation (NMT) systems.One plausible explanation for this discrepancy is that the straightforward prompts employed in these methodologies are unable to fully exploit the acquired instruction-following capabilities.To this end, we propose the TASTE framework, which stands for translating through selfreflection.The self-reflection process includes two stages of inference.In the first stage, LLMs are instructed to generate preliminary translations and conduct self-assessments on these translations simultaneously.In the second stage, LLMs are tasked to refine these preliminary translations according to the evaluation results.The evaluation results in four language directions on the WMT22 benchmark reveal the effectiveness of our approach compared to existing methods.Our work presents a promising approach to unleash the potential of LLMs and enhance their capabilities in MT.The codes and datasets are open-sourced at https://github. com/YutongWang1216/ReflectionLLMMT.

This work introduces an original and practical paradigm for narrative comprehension, stemming from the characteristics that individual passages within narratives tend to be more cohesively related than isolated.Complementary to the common end-to-end paradigm, we propose a fine-grained modeling of narrative context, by formulating a graph dubbed NARCO, which explicitly depicts task-agnostic coherence dependencies that are ready to be consumed by various downstream tasks.In particular, edges in NARCO encompass free-form retrospective questions between context snippets, inspired by human cognitive perception that constantly reinstates relevant events from prior context.Importantly, our graph formalism is practically instantiated by LLMs without human annotations, through our designed two-stage prompting scheme.To examine the graph properties and its utility, we conduct three studies in narratives, each from a unique angle: edge relation efficacy, local context enrichment, and broader application in QA.All tasks could benefit from the explicit coherence captured by NARCO.

Retrieval-augmented generation (RAG) enhances large language models (LLMs) by incorporating additional information from retrieval.However, studies have shown that LLMs still face challenges in effectively using the retrieved information, even ignoring it or being misled by it.The key reason is that the training of LLMs does not clearly make LLMs learn how to utilize input retrieved texts with varied quality.In this paper, we propose a novel perspective that considers the role of LLMs in RAG as "Information Refiner", which means that regardless of correctness, completeness, or usefulness of retrieved texts, LLMs can consistently integrate knowledge within the retrieved texts and model parameters to generate the texts that are more concise, accurate, and complete than the retrieved texts.To this end, we propose an information refinement training method named INFO-RAG that optimizes LLMs for RAG in an unsupervised manner.INFO-RAG is low-cost and general across various tasks.Extensive experiments on zero-shot prediction of 11 datasets in diverse tasks including Question Answering, Slot-Filling, Language Modeling, Dialogue, and Code Generation show that INFO-RAG improves the performance of LLaMA2 by an average of 9.39% relative points.INFO-RAG also shows advantages in in-context learning and robustness of RAG.LLMs

Prior research in Image Machine Translation (IMT) has focused on either translating the source image solely into the target language text or exclusively into the target image. As a result, the former approach lacked the capacity to generate target images, while the latter was insufficient in producing target text. In this paper, we present a Unified Multimodal Text Image Translation (UMTIT) model that not only translates text images into the target language but also generates consistent target images. The UMTIT model consists of two image-text modality conversion steps: the first step converts images to text to recognize the source text and generate translations, while the second step transforms text to images to create target images based on the translations. Due to the limited availability of public datasets, we have constructed two multimodal image translation datasets. Experimental results show that our UMTIT model is versatile enough to handle tasks across multiple modalities and outperforms previous methods. Notably, UMTIT surpasses the state-of-the-art TrOCR in text recognition tasks, achieving a lower Character Error Rate (CER); it also outperforms cascading methods in text translation tasks, obtaining a higher BLEU score; and, most importantly, UMTIT can generate high-quality target text images.

Minimum Bayesian Risk Decoding (MBR) emerges as a promising decoding algorithm in Neural Machine Translation. However, MBR performs poorly with label smoothing, which is surprising as label smoothing provides decent improvement with beam search and improves generality in various tasks. In this work, we show that the issue arises from the inconsistency of label smoothing on the token-level and sequence-level distributions. We demonstrate that even though label smoothing only causes a slight change in the token level, the sequence-level distribution is highly skewed. We coin the issue autoregressive over-smoothness. To address this issue, we propose a simple and effective method, Distributional Cooling MBR (DC-MBR), which manipulates the entropy of output distributions by tuning down the Softmax temperature. We theoretically prove the equivalence between the pre-tuning label smoothing factor and distributional cooling. Extensive experiments on NMT benchmarks validate that distributional cooling improves MBR in various settings.

Yiju Guo, Ganqu Cui, Lifan Yuan, Ning Ding, Zexu Sun, Bowen Sun, Huimin Chen, Ruobing Xie, Jie Zhou, Yankai Lin, Zhiyuan Liu, Maosong Sun. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

Chinese Spell Checking (CSC) aims to detect and correct spelling errors in sentences.Despite Large Language Models (LLMs) exhibit robust capabilities and are widely applied in various tasks, their performance on CSC is often unsatisfactory.We find that LLMs fail to meet the Chinese character-level constraints of the CSC task, namely equal length and phonetic similarity, leading to a performance bottleneck.Further analysis reveals that this issue stems from the granularity of tokenization, as current mixed character-word tokenization struggles to satisfy these characterlevel constraints.To address this issue, we propose C-LLM, a Large Language Modelbased Chinese Spell Checking method that learns to check errors Character by Character.Character-level tokenization enables the model to learn character-level alignment, effectively mitigating issues related to character-level constraints.Furthermore, CSC is simplified to replication-dominated and substitutionsupplemented tasks.Experiments on two CSC benchmarks demonstrate that C-LLM achieves an average improvement of 10% over existing methods.Specifically, it shows a 2.1% improvement in general scenarios and a significant 12% improvement in vertical domain scenarios, establishing state-of-the-art performance.The source code can be accessed at https://github.com/ktlKTL/C-LLM.

Liang Zhang, Zhen Yang, Biao Fu, Ziyao Lu, Liangying Shao, Shiyu Liu, Fandong Meng, Jie Zhou, Xiaoli Wang, Jinsong Su. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

Note-level automatic singing transcription, involving the extraction of onset, offset, and pitch information from a singing voice, is a crucial process in the field of Music Information Retrieval (MIR), The recent advancements in deep learning models have led to significant progress in this field. However, annotating a training dataset requires professional music expertise, and the entire annotation process is time-consuming and labor-intensive. Therefore, this field suffers from a severe data scarcity problem. To address this issue, we developed a singing transcription model based on wav2vec 2.0, a pretrained speech representation model. The model can learn from unlabeled speech and weakly-labeled singing data and use this knowledge to benefit the transcription task. The experiments showed that our proposed method achieves a significant improvement over previous approaches on various benchmarks. Moreover, additional experiments demonstrate that our method achieves competitive performance even with a small proportion of training data.

Autonomous agents empowered by Large Language Models (LLMs) have undergone significant improvements, enabling them to generalize across a broad spectrum of tasks. However, in real-world scenarios, cooperation among individuals is often required to enhance the efficiency and effectiveness of task accomplishment. Hence, inspired by human group dynamics, we propose a multi-agent framework AgentVerse that can effectively orchestrate a collaborative group of expert agents as a greater-than-the-sum-of-its-parts system. Our experiments demonstrate that AgentVerse can proficiently deploy multi-agent groups that outperform a single agent. Extensive experiments on text understanding, reasoning, coding, tool utilization, and embodied AI confirm the effectiveness of AgentVerse. Moreover, our analysis of agent interactions within AgentVerse reveals the emergence of specific collaborative behaviors, contributing to heightened group efficiency. We will release our codebase, AgentVerse, to further facilitate multi-agent research.

Despite the advancements of open-source large language models (LLMs), e.g., LLaMA, they remain significantly limited in tool-use capabilities, i.e., using external tools (APIs) to fulfill human instructions. The reason is that current instruction tuning largely focuses on basic language tasks but ignores the tool-use domain. This is in contrast to the excellent tool-use capabilities of state-of-the-art (SOTA) closed-source LLMs, e.g., ChatGPT. To bridge this gap, we introduce ToolLLM, a general tool-use framework encompassing data construction, model training, and evaluation. We first present ToolBench, an instruction-tuning dataset for tool use, which is constructed automatically using ChatGPT. Specifically, the construction can be divided into three stages: (i) API collection: we collect 16,464 real-world RESTful APIs spanning 49 categories from RapidAPI Hub; (ii) instruction generation: we prompt ChatGPT to generate diverse instructions involving these APIs, covering both single-tool and multi-tool scenarios; (iii) solution path annotation: we use ChatGPT to search for a valid solution path (chain of API calls) for each instruction. To enhance the reasoning capabilities of LLMs, we develop a novel depth-first search-based decision tree algorithm. It enables LLMs to evaluate multiple reasoning traces and expand the search space. Moreover, to evaluate the tool-use capabilities of LLMs, we develop an automatic evaluator: ToolEval. Based on ToolBench, we fine-tune LLaMA to obtain an LLM ToolLLaMA, and equip it with a neural API retriever to recommend appropriate APIs for each instruction. Experiments show that ToolLLaMA demonstrates a remarkable ability to execute complex instructions and generalize to unseen APIs, and exhibits comparable performance to ChatGPT. Our ToolLLaMA also demonstrates strong zero-shot generalization ability in an out-of-distribution tool-use dataset: APIBench.

As large language models (LLMs) generate texts with increasing fluency and realism, there is a growing need to identify the source of texts to prevent the abuse of LLMs. Text watermarking techniques have proven reliable in distinguishing whether a text is generated by LLMs by injecting hidden patterns. However, we argue that existing LLM watermarking methods are encoding-inefficient and cannot flexibly meet the diverse information encoding needs (such as encoding model version, generation time, user id, etc.). In this work, we conduct the first systematic study on the topic of **Codable Text Watermarking for LLMs** (CTWL) that allows text watermarks to carry multi-bit customizable information. First of all, we study the taxonomy of LLM watermarking technologies and give a mathematical formulation for CTWL. Additionally, we provide a comprehensive evaluation system for CTWL: (1) watermarking success rate, (2) robustness against various corruptions, (3) coding rate of payload information, (4) encoding and decoding efficiency, (5) impacts on the quality of the generated text. To meet the requirements of these non-Pareto-improving metrics, we follow the most prominent vocabulary partition-based watermarking direction, and devise an advanced CTWL method named **Balance-Marking**. The core idea of our method is to use a proxy language model to split the vocabulary into probability-balanced parts, thereby effectively maintaining the quality of the watermarked text. Our code is available at https://github.com/lancopku/codable-watermarking-for-llm.

Multiple choice questions (MCQs) serve as a common yet important task format in the evaluation of large language models (LLMs). This work shows that modern LLMs are vulnerable to option position changes in MCQs due to their inherent “selection bias”, namely, they prefer to select specific option IDs as answers (like “Option A”). Through extensive empirical analyses with 20 LLMs on three benchmarks, we pinpoint that this behavioral bias primarily stems from LLMs’ token bias, where the model a priori assigns more probabilistic mass to specific option ID tokens (e.g., A/B/C/D) when predicting answers from the option IDs. To mitigate selection bias, we propose a label-free, inference-time debiasing method, called PriDe, which separates the model’s prior bias for option IDs from the overall prediction distribution. PriDe first estimates the prior by permutating option contents on a small number of test samples, and then applies the estimated prior to debias the remaining samples. We demonstrate that it achieves interpretable and transferable debiasing with high computational efficiency. We hope this work can draw broader research attention to the bias and robustness of modern LLMs.

Language generation based on maximum likelihood estimation (MLE) has become the fundamental approach for text generation. Maximum likelihood estimation is typically performed by minimizing the log-likelihood loss, also known as the logarithmic score in statistical decision theory. The logarithmic score is strictly proper in the sense that it encourages honest forecasts, where the expected score is maximized only when the model reports true probabilities. Although many strictly proper scoring rules exist, the logarithmic score is the only local scoring rule among them that depends exclusively on the probability of the observed sample, making it capable of handling the exponentially large sample space of natural text. In this work, we propose a straightforward strategy for adapting scoring rules to language generation, allowing for language modeling with any non-local scoring rules. Leveraging this strategy, we train language generation models using two classic strictly proper scoring rules, the Brier score and the Spherical score, as alternatives to the logarithmic score. Experimental results indicate that simply substituting the loss function, without adjusting other hyperparameters, can yield substantial improvements in model’s generation capabilities. Moreover, these improvements can scale up to large language models (LLMs) such as LLaMA-7B and LLaMA-13B. Source code: https://github.com/shaochenze/ScoringRulesLM.

When reading a story, humans can quickly understand new fictional characters with a few observations, mainly by drawing analogies to fictional and real people they already know. This reflects the few-shot and meta-learning essence of humans' inference of characters' mental states, *i.e.*, theory-of-mind (ToM), which is largely ignored in existing research. We fill this gap with a novel NLP dataset in a realistic narrative understanding scenario, ToM-in-AMC. Our dataset consists of $\sim$1,000 parsed movie scripts, each corresponding to a few-shot character understanding task that requires models to mimic humans' ability of fast digesting characters with a few starting scenes in a new movie. We further propose a novel ToM prompting approach designed to explicitly assess the influence of multiple ToM dimensions. It surpasses existing baseline models, underscoring the significance of modeling multiple ToM dimensions for our task. Our extensive human study verifies that humans are capable of solving our problem by inferring characters' mental states based on their previously seen movies. In comparison, all the AI systems lag $>20\%$ behind humans, highlighting a notable limitation in existing approaches' ToM capabilities. Code and data are available at https://github.com/ShunchiZhang/ToM-in-AMC

Pre-trained Transformers inherently possess the characteristic of sparse activation, where only a small fraction of the neurons are activated for each token. While sparse activation has been explored through post-training methods, its potential in pre-training remains untapped. In this work, we first study how activation properties change during pre-training. Our examination reveals that Transformers exhibit sparse activation throughout the majority of the pre-training process while the activation correlation keeps evolving as training progresses. Leveraging this observation, we propose Switchable Sparse-Dense Learning (SSD). SSD adaptively switches between the Mixtures-of-Experts (MoE) based sparse training and the conventional dense training during the pre-training process, leveraging the efficiency of sparse training and avoiding the static activation correlation of sparse training. Compared to dense training, SSD achieves comparable performance with identical model size and reduces pre-training costs. Moreover, the models trained with SSD can be directly used as MoE models for sparse inference and achieve the same performance as dense models with up to $2\times$ faster inference speed. Codes are available at https://github.com/thunlp/moefication.

Prepending model inputs with safety prompts is a common practice for safeguarding large language models (LLMs) against queries with harmful intents. However, the underlying working mechanisms of safety prompts have not been unraveled yet, restricting the possibility of automatically optimizing them to improve LLM safety. In this work, we investigate how LLMs’ behavior (i.e., complying with or refusing user queries) is affected by safety prompts from the perspective of model representation. We find that in the representation space, the input queries are typically moved by safety prompts in a "higher-refusal" direction, in which models become more prone to refusing to provide assistance, even when the queries are harmless. On the other hand, LLMs are naturally capable of distinguishing harmful and harmless queries without safety prompts. Inspired by these findings, we propose a method for safety prompt optimization, namely DRO (Directed Representation Optimization). Treating a safety prompt as continuous, trainable embeddings, DRO learns to move the queries’ representations along or opposite the refusal direction, depending on their harmfulness. Experiments with eight LLMs on out-of-domain and jailbreak benchmarks demonstrate that DRO remarkably improves the safeguarding performance of human-crafted safety prompts, without compromising the models’ general performance.

Code language models such as CodeT5 and CodeLlama have demonstrated substantial achievement in code comprehension. While the majority of research efforts have focused on improving model architectures and training processes, we find that the current benchmarks used for evaluating code comprehension models are confined to high-readability code, regardless of the popularity of low-readability code in reality. As such, they are inadequate to demonstrate the full spectrum of the model's ability, particularly the robustness to varying readability degrees. In this paper, we analyze the robustness of code summarization models to code with varying readability, including seven obfuscated datasets derived from existing benchmarks. Our findings indicate that current code summarization models are vulnerable to code with poor readability. In particular, their performance predominantly depends on semantic cues within the code, often neglecting the syntactic aspects. Existing benchmarks are biased toward evaluating semantic features, thereby overlooking the models' ability to understand nonsensitive syntactic features. Based on the findings, we present Poor-CodeSumEval, a new evaluation benchmark on code summarization tasks. PoorCodeSumEval innovatively introduces readability into the testing process, considering semantic, syntactic, and their cross-obfuscation, thereby providing a more comprehensive and rigorous evaluation of code summarization models. Our studies also provide more insightful suggestions for future research, such as constructing multi-readability benchmarks to evaluate the robustness of models on poor-readability code, proposing readability-awareness metrics, and automatic methods for code data cleaning and normalization.

Che Jiang, Biqing Qi, Xiangyu Hong, Dayuan Fu, Yang Cheng, Fandong Meng, Mo Yu, Bowen Zhou, Jie Zhou. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Yun Luo, Zhen Yang, Fandong Meng, Yingjie Li, Fang Guo, Qinglin Qi, Jie Zhou, Yue Zhang. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Driven by the rapid development of Large Language Models (LLMs), LLM-based agents have been developed to handle various real-world applications, including finance, healthcare, and shopping, etc. It is crucial to ensure the reliability and security of LLM-based agents during applications. However, the safety issues of LLM-based agents are currently under-explored. In this work, we take the first step to investigate one of the typical safety threats, backdoor attack, to LLM-based agents. We first formulate a general framework of agent backdoor attacks, then we present a thorough analysis of different forms of agent backdoor attacks. Specifically, compared with traditional backdoor attacks on LLMs that are only able to manipulate the user inputs and model outputs, agent backdoor attacks exhibit more diverse and covert forms: (1) From the perspective of the final attacking outcomes, the agent backdoor attacker can not only choose to manipulate the final output distribution, but also introduce the malicious behavior in an intermediate reasoning step only, while keeping the final output correct. (2) Furthermore, the former category can be divided into two subcategories based on trigger locations, in which the backdoor trigger can either be hidden in the user query or appear in an intermediate observation returned by the external environment. We implement the above variations of agent backdoor attacks on two typical agent tasks including web shopping and tool utilization. Extensive experiments show that LLM-based agents suffer severely from backdoor attacks and such backdoor vulnerability cannot be easily mitigated by current textual backdoor defense algorithms. This indicates an urgent need for further research on the development of targeted defenses against backdoor attacks on LLM-based agents. Warning: This paper may contain biased content.

In recent years, we have witnessed significant improvements in pre-trained language models (PLM) brought about by the scaling of parameter sizes and data amounts. However, this also brings high computational and storage costs. In this paper, we present a new direction to improve PLMs without scaling parameters and data: adopting a geometric feature space that is more suitable for encoding the intrinsic structured features of text. Although text is generally considered unstructured data, it possesses rich intrinsic structured features that signify syntactic and semantic relationships. Leveraging these structured features is vital for text understanding. Given that structured features are better encoded in hyperbolic spaces than in the Euclidean spaces used by conventional PLMs, we propose that PLMs should operate entirely within hyperbolic spaces. Our experiments demonstrate the superiority of hyperbolic PLMs over Euclidean PLMs across a wide variety of tasks, using the same parameter and data settings. This suggests that altering the geometry of model representation is a promising direction for model enhancement. The code is released athttps://github.com/thunlp/hyperbolic_llm

Why can pre-trained language models (PLMs) learn universal representations and effectively adapt to broad NLP tasks differing a lot superficially? In this work, we empirically find evidence indicating that the adaptations of PLMs to various few-shot tasks can be reparameterized as optimizing only a few free parameters in a unified low-dimensionalintrinsic task subspace, which may help us understand why PLMs could easily adapt to various NLP tasks with small-scale data. To find such a subspace and examine its universality, we propose an analysis pipeline calledintrinsic prompt tuning(IPT). Specifically, we resort to the recent success of prompt tuning and decompose the soft prompts of multiple NLP tasks into the same low-dimensional nonlinear subspace, then we learn to adapt the PLM to unseen data or tasks by only tuning parameters in this subspace. In the experiments, we study diverse few-shot NLP tasks and surprisingly find that in a 250-dimensional subspace found with 100 tasks, by only tuning 250 free parameters, we can recover 97% and 83% of the full prompt tuning performance for 100 seen tasks (using different training data) and 20 unseen tasks, respectively, showing great generalization ability of the found intrinsic task subspace. Besides being an analysis tool, IPTcould further help us improve the prompt tuning stability.

This article introduces the Tenth Dialog System Technology Challenge (DSTC-10). This edition of the DSTC focuses on applying end-to-end dialog technologies for five distinct tasks in dialog systems, namely 1. Incorporation of Meme images into open domain dialogs, 2. Knowledge-grounded Task-oriented Dialogue Modeling on Spoken Conversations, 3. Situated Interactive Multimodal dialogs, 4. Reasoning for Audio Visual Scene-Aware Dialog, and 5. Automatic Evaluation and Moderation of Open-domainDialogue Systems. This article describes the task definition, provided datasets, baselines, and evaluation setup for each track. We also summarize the results of the submitted systems to highlight the general trends of the state-of-the-art technologies for the tasks.

Zero-shot information extraction (IE) tasks have attracted great attention recently. However, how to jointly model multiple IE tasks in the zero-shot scenario is still an open question. In this article, we focus on zero-shot joint IE tasks and highlight how to transfer the knowledge of cross-task relations from the source domain to the target domain. To solve this problem, we first unify all IE tasks with a machine reading comprehension (MRC) framework, which can make the most of training data and enhance its ability on span extraction. Then, we generatecomplex questionsto explicitly model cross-task relations with natural language descriptions, thereby providing prior knowledge for pre-defined types and building more general linkages among different entities and triggers as well. Specifically, we define three operations for generating templates for complex questions, i.e.,intersecting,connecting, andcomposing. Besides, we design an efficient training strategy to exploit the synthetic data with complex questions. We evaluate our approach on four datasets from different domains for various IE tasks. Experimental results show the effectiveness of our approach in improving the performance of zero-shot joint IE tasks in multiple domains.

Cross-domain recommendation (CDR) aims at leveraging the correlation of users’ behaviors in both the source and target domains to improve the user preference modeling in the target domain. Conventional CDR methods typically explore the dual-relations between the source and target domains’ behaviors. However, this may ignore the informative mixed behaviors that naturally reflect the user’s global preference. To address this issue, we present a novel framework, termed triple sequence learning for cross-domain recommendation (Tri-CDR), which jointly models the source, target, and mixed behavior sequences to highlight the global and target preference and precisely model the triple correlation in CDR. Specifically, Tri-CDR independently models the hidden representations for the triple behavior sequences and proposes a triple cross-domain attention (TCA) method to emphasize the informative knowledge related to both user’s global and target-domain preference. To comprehensively explore the cross-domain correlations, we design a triple contrastive learning (TCL) strategy that simultaneously considers the coarse-grained similarities and fine-grained distinctions among the triple sequences, ensuring the alignment while preserving information diversity in multi-domain. We conduct extensive experiments and analyses on six cross-domain settings. The significant improvements of Tri-CDR with different sequential encoders verify its effectiveness and universality. The source code is available at https://github.com/hulkima/Tri-CDR .

Non-autoregressive neural machine translation (NAT) models suffer from the multi-modality problem that there may exist multiple possible translations of a source sentence, so the reference sentence may be inappropriate for the training when the NAT output is closer to other translations. In response to this problem, we introduce a rephraser to provide a better training target for NAT by rephrasing the reference sentence according to the NAT output. As we train NAT based on the rephraser output rather than the reference sentence, the rephraser output should fit well with the NAT output and not deviate too far from the reference, which can be quantified as reward functions and optimized by reinforcement learning. Experiments on major WMT benchmarks and NAT baselines show that our approach consistently improves the translation quality of NAT. Specifically, our best variant achieves comparable performance to the autoregressive Transformer, while being 14.7 times more efficient in inference.

Multimodal abstractive summarization (MAS) aims to produce a concise summary given the multimodal data (text and vision).Existing studies mainly focus on how to effectively use the visual features from the perspective of an article, having achieved impressive success on the high-resource English dataset.However, less attention has been paid to the visual features from the perspective of the summary, which may limit the model performance, especially in the low-and zero-resource scenarios.In this paper, we propose to improve the summary quality through summary-oriented visual features.To this end, we devise two auxiliary tasks including vision to summary task and masked image modeling task.Together with the main summarization task, we optimize the MAS model via the training objectives of all these tasks.By these means, the MAS model can be enhanced by capturing the summaryoriented visual features, thereby yielding more accurate summaries.Experiments on 44 languages, covering mid-high-, low-, and zeroresource scenarios, verify the effectiveness and superiority of the proposed approach, which achieves state-of-the-art performance under all scenarios.Additionally, we will contribute a large-scale multilingual multimodal abstractive summarization (MM-Sum) dataset. 1

Jiaan Wang, Fandong Meng, Duo Zheng, Yunlong Liang, Zhixu Li, Jianfeng Qu, Jie Zhou. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Existing neural models have difficulty generalizing to unseen combinations of seen components.To achieve compositional generalization, models are required to consistently interpret (sub)expressions across contexts.Without modifying model architectures, we improve the capability of Transformer on compositional generalization through consistency regularization training, which promotes representation consistency across samples and prediction consistency for a single sample.Experimental results on semantic parsing and machine translation benchmarks empirically demonstrate the effectiveness and generality of our method.In addition, we find that the prediction consistency scores on in-distribution validation sets can be an alternative for evaluating models during training, when commonly-used metrics are not informative.

Mo Yu, Jiangnan Li, Shunyu Yao, Wenjie Pang, Xiaochen Zhou, Zhou Xiao, Fandong Meng, Jie Zhou. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Jiali Zeng, Yufan Jiang, Yongjing Yin, Yi Jing, Fandong Meng, Binghuai Lin, Yunbo Cao, Jie Zhou. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Zhengyan Zhang, Zhiyuan Zeng, Yankai Lin, Huadong Wang, Deming Ye, Chaojun Xiao, Xu Han, Zhiyuan Liu, Peng Li, Maosong Sun, Jie Zhou. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Dialogue comprehension and generation are vital to the success of open-domain dialogue systems.Although pre-trained generative conversation models have made significant progress in generating fluent responses, people have difficulty judging whether they understand and efficiently model the contextual information of the conversation.In this study, we propose a Multi-Source Probing (MSP) method to probe the dialogue comprehension abilities of opendomain dialogue models.MSP aggregates features from multiple sources to accomplish diverse task goals and conducts downstream tasks in a generative manner that is consistent with dialogue model pre-training to leverage model capabilities.We conduct probing experiments on seven tasks that require various dialogue comprehension skills, based on the internal representations encoded by dialogue models.Experimental results show that open-domain dialogue models can encode semantic information in the intermediate hidden states, which facilitates dialogue comprehension tasks.Models of different scales and structures possess different conversational understanding capabilities.Our findings encourage a comprehensive evaluation and design of open-domain dialogue models.

In-context learning (ICL) emerges as a promising capability of large language models (LLMs) by providing them with demonstration examples to perform diverse tasks.However, the underlying mechanism of how LLMs learn from the provided context remains under-explored.In this paper, we investigate the working mechanism of ICL through an information flow lens.Our findings reveal that label words in the demonstration examples function as anchors:(1) semantic information aggregates into label word representations during the shallow computation layers' processing; (2) the consolidated information in label words serves as a reference for LLMs' final predictions.Based on these insights, we introduce an anchor re-weighting method to improve ICL performance, a demonstration compression technique to expedite inference, and an analysis framework for diagnosing ICL errors in GPT2-XL.The promising applications of our findings again validate the uncovered ICL working mechanism and pave the way for future studies. 1

Pre-training on large-scale open-domain dialogue data can substantially improve the performance of dialogue models.However, the pre-trained dialogue model's ability to utilize long-range context is limited due to the scarcity of long-turn dialogue sessions.Most dialogues in existing pre-training corpora contain fewer than three turns of dialogue.To alleviate this issue, we propose the Retrieve, Reorganize and Rescale framework (Re 3 Dial), which can automatically construct billion-scale long-turn dialogues by reorganizing existing short-turn ones.Given a short-turn session, Re 3 Dial first employs a session retriever to retrieve coherent consecutive sessions.To this end, we train the retriever to capture semantic and discourse relations within multi-turn dialogues through contrastive training.Next, Re 3 Dial samples a session from retrieved results following a diversity sampling strategy, which is designed to penalize repetitive or generic sessions.A longer session is then derived by concatenating the original session and the sampled session.By repeating the above process, Re 3 Dial can yield a coherent long-turn dialogue.Extensive experiments on multiple multi-turn dialogue benchmarks demonstrate that Re 3 Dial significantly improves the dialogue model's ability to utilize long-range context and thus generate more sensible and informative responses.Finally, we build a toolkit for efficiently rescaling conversations with Re 3 Dial, which enables us to construct a corpus containing 1B Chinese dialogue sessions with 11.3 turns on average (5× longer than the original corpus).Our retriever model, code, and data is publicly available at https://github.com/thu-coai/Re3Dial.

Few-shot relation extraction (FSRE) aims to train a model that can deal with new relations using only a few labeled examples.Most existing studies employ Prototypical Networks for FSRE, which usually overfits the relation classes in the training set and cannot generalize well to unseen relations.By investigating the class separation of an FSRE model, we find that model upper layers are prone to learn relation-specific knowledge.Therefore, in this paper, we propose a HyperNetworkbased Decoupling approach to improve the generalization of FSRE models.Specifically, our model consists of an encoder, a network generator (for producing relation classifiers) and the generated-then-finetuned classifiers for every N -way-K-shot episode.Meanwhile, we design a two-step training strategy along with a class-agnostic aligner, by which the generated classifiers focus on acquiring relation-specific knowledge and the encoder is encouraged to learn more general relation knowledge.In this way, the roles of upper and lower layers in our FSRE model are explicitly decoupled, thus enhancing its generalizing capability during testing.Experiments on two public datasets demonstrate the effectiveness of our method.Our source code is available at https: //github.com/DeepLearnXMU/FSRE-HDN.

Creating a piece of music is difficult for people who have never been trained to compose. We present an automatic music generation system to lower the threshold of creating music. The system takes the user's humming as input and creates full music based on the humming melody. The system consists of five modules: 1) humming transcription, 2) melody generation, 3) broken chord generation, 4) accompaniment generation, and 5) audio synthesis. The first module transcribes the user's humming audio to a score, and then the melody generation module composes a complete melody based on the user's humming melody. After that, the third module will generate a broken chord track to accompany the full melody, and the fourth module will create more accompanying tracks. Finally, the audio synthesis module mixes all the tracks to generate the music. Through the user experiment, our system can generate high-quality music with natural expression based on the user's humming input.

This paper presents a novel system, named LingGe ("伶歌" in Chinese), to generate songs for ancient Chinese poems automatically. LingGe takes the poem as the lyric, composes music conditioned on the lyric, and finally outputs a full song including the singing and the accompaniment. It consists of four modules: rhythm recognition, melody generation, accompaniment generation, and audio synthesis. Firstly, the rhythm recognition module analyzes the song structure and rhythm according to the poem. Secondly, the melody generation module assembles the rhythm into the template and then generates the melody. Thirdly, the accompaniment generation module predicts the accompaniment in harmony with the melody. Finally, the audio synthesis module generates singing and accompaniment audio and then mixes them to obtain songs. The results show that LingGe can generate high-quality and expressive songs for ancient Chinese poems, both in harmony and rhythm.

Multi-modal keyphrase generation aims to produce a set of keyphrases that represent the core points of the input text-image pair. In this regard, dominant methods mainly focus on multi-modal fusion for keyphrase generation. Nevertheless, there are still two main drawbacks: 1) only a limited number of sources, such as image captions, can be utilized to provide auxiliary information. However, they may not be sufficient for the subsequent keyphrase generation. 2) the input text and image are often not perfectly matched, and thus the image may introduce noise into the model. To address these limitations, in this paper, we propose a novel multi-modal keyphrase generation model, which not only enriches the model input with external knowledge, but also effectively filters image noise. First, we introduce external visual entities of the image as the supplementary input to the model, which benefits the cross-modal semantic alignment for keyphrase generation. Second, we simultaneously calculate an image-text matching score and image region-text correlation scores to perform multi-granularity image noise filtering. Particularly, we introduce the correlation scores between image regions and ground-truth keyphrases to refine the calculation of the previously-mentioned correlation scores. To demonstrate the effectiveness of our model, we conduct several groups of experiments on the benchmark dataset. Experimental results and in-depth analyses show that our model achieves the state-of-the-art performance. Our code is available on https://github.com/DeepLearnXMU/MM-MKP.

Federated learning algorithms enable neural network models to be trained across multiple decentralized edge devices without sharing private data. However, they are susceptible to backdoor attacks launched by malicious clients. Existing robust federated aggregation algorithms heuristically detect and exclude suspicious clients based on their parameter distances, but they are ineffective on Natural Language Processing (NLP) tasks. The main reason is that, although text backdoor patterns are obvious at the underlying dataset level, they are usually hidden at the parameter level, since injecting backdoors into texts with discrete feature space has less impact on the statistics of the model parameters. To settle this issue, we propose to identify backdoor clients by explicitly modeling the data divergence among clients in federated NLP systems. Through theoretical analysis, we derive the f-divergence indicator to estimate the client data divergence with aggregation updates and Hessians. Furthermore, we devise a dataset synthesization method with a Hessian reassignment mechanism guided by the diffusion theory to address the key challenge of inaccessible datasets in calculating clients' data Hessians. We then present the novel Federated F-Divergence-Based Aggregation~(\textbf{Fed-FA}) algorithm, which leverages the f-divergence indicator to detect and discard suspicious clients. Extensive empirical results show that Fed-FA outperforms all the parameter distance-based methods in defending against backdoor attacks among various natural language backdoor attack scenarios.

Recommendation usually focuses on immediate accuracy metrics like CTR as training objectives. User retention rate, which reflects the percentage of today’s users that will return to the recommender system in the next few days, should be paid more attention to in real-world systems. User retention is the most intuitive and accurate reflection of user long-term satisfaction. However, most existing recommender systems are not focused on user retention-related objectives, since their complexity and uncertainty make it extremely hard to discover why a user will or will not return to a system and which behaviors affect user retention. In this work, we conduct a series of preliminary explorations on discovering and making full use of the reasons for user retention in recommendation. Specifically, we make a first attempt to design a rationale contrastive multi-instance learning framework to explore the rationale and improve the interpretability of user retention. Extensive offline and online evaluations with detailed analyses of a real-world recommender system verify the effectiveness of our user retention modeling. We further reveal the real-world interpretable factors of user retention from both user surveys and explicit negative feedback quantitative analyses to facilitate future model designs. The source codes are released at https://github.com/dinry/IURO.

Negative Sampling in recommendation aims to capture informative negative instances for the sparse user-item interactions to improve the performance. Conventional negative sampling methods tend to select informative hard negative samples (HNS) besides the default random samples. However, these hard negative sampling methods usually struggle with false hard negative samples (FHNS), which happens when a user-item interaction has not been observed yet and is picked as a negative sample, while the user will actually interact with this item once exposed to it. Such FHNS issues may seriously confuse the model training, while most conventional hard negative sampling methods do not systematically explore and distinguish FHNS from HNS. To address this issue, we propose a novel model-agnostic Real Hard Negative Sampling (RealHNS) framework specially for cross-domain recommendation (CDR), which aims to discover the false and refine the real from all HNS via both general and cross-domain real hard negative sample selectors. For the general part, we conduct the coarse- and fine-grained real HNS selectors sequentially, armed with a dynamic item-based FHNS filter to find high-quality HNS. For the cross-domain part, we further design a new cross-domain HNS for alleviating negative transfer in CDR and discover its corresponding FHNS via a dynamic user-based FHNS filter to keep its power. We conduct experiments on four datasets based on three representative hard negative sampling methods, along with extensive model analyses, ablation studies, and universality analyses. The consistent improvements indicate the effectiveness, robustness, and universality of RealHNS, which is also easy-to-deploy in real-world systems as a plug-and-play strategy. The source code is avaliable in https://github.com/hulkima/RealHNS.

Recently, a series of pioneer studies have shown the potency of pre-trained models in sequential recommendation, illuminating the path of building an omniscient unified pre-trained recommendation model for different downstream recommendation tasks. Despite these advancements, the vulnerabilities of classical recommender systems also exist in pre-trained recommendation in a new form, while the security of pre-trained recommendation model is still unexplored, which may threaten its widely practical applications. In this study, we propose a novel framework for backdoor attacking in pre-trained recommendation. We demonstrate the provider of the pre-trained model can easily insert a backdoor in pre-training, thereby increasing the exposure rates of target items to target user groups. Specifically, we design two novel and effective backdoor attacks: basic replacement and prompt-enhanced, under various recommendation pre-training usage scenarios. Experimental results on real-world datasets show that our proposed attack strategies significantly improve the exposure rates of target items to target users by hundreds of times in comparison to the clean model. The source codes are released in https://github.com/wyqing20/APRec.

Neural chat translation (NCT) aims to translate a cross-lingual chat between speakers of different languages. Existing context-aware NMT models cannot achieve satisfactory performances due to the following inherent problems: 1) limited resources of annotated bilingual dialogues; 2) the neglect of modelling conversational properties; 3) training discrepancy between different stages. To address these issues, in this paper, we propose a multi-task multi-stage transitional (MMT) training framework, where an NCT model is trained using the bilingual chat translation dataset and additional monolingual dialogues. We elaborately design two auxiliary tasks, namely utterance discrimination and speaker discrimination, to introduce the modelling of dialogue coherence and speaker characteristic into the NCT model. The training process consists of three stages: 1) sentence-level pre-training on large-scale parallel corpus; 2) intermediate training with auxiliary tasks using additional monolingual dialogues; 3) context-aware fine-tuning with gradual transition. Particularly, the second stage serves as an intermediate phase that alleviates the training discrepancy between the pre-training and fine-tuning stages. Moreover, to make the stage transition smoother, we train the NCT model using a gradual transition strategy, i.e., gradually transiting from using monolingual to bilingual dialogues. Extensive experiments on two language pairs demonstrate the effectiveness and superiority of our proposed training framework.

The pre-trained model (PTM) is revolutionizing Artificial Intelligence (AI) technology. However, the hardware requirement of PTM training is prohibitively high, making it a game for a small proportion of people. Therefore, we proposed PatrickStar system to lower the hardware requirements of PTMs and make them accessible to everyone. PatrickStar uses the CPU-GPU heterogeneous memory space to store the model data. Different from existing works, we organize the model data in memory chunks and dynamically distribute them in the heterogeneous memory. Guided by the runtime memory statistics collected in a warm-up iteration, chunks are orchestrated efficiently in heterogeneous memory and generate lower CPU-GPU data transmission volume and higher bandwidth utilization. Symbiosis with the Zero Redundancy Optimizer, PatrickStar scales to multiple GPUs on multiple nodes. The system can train tasks on bigger models and larger batch sizes, which cannot be accomplished by existing works. Experimental results show that PatrickStar extends model scales 2.27 and 2.5 times of DeepSpeed, and exhibits significantly higher execution speed. PatricStar also successfully runs the 175B GPT3 training task on a 32 GPU cluster. Our code is available athttps://github.com/Tencent/PatrickStar.

Weize Chen, Xu Han, Yankai Lin, Hexu Zhao, Zhiyuan Liu, Peng Li, Maosong Sun, Jie Zhou. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.

Existing reference-free metrics have obvious limitations for evaluating controlled text generation models.Unsupervised metrics can only provide a task-agnostic evaluation result which correlates weakly with human judgments, whereas supervised ones may overfit task-specific data with poor generalization ability to other datasets.In this paper, we propose an unsupervised reference-free metric called CTRLEval, which evaluates controlled text generation from different aspects by formulating each aspect into multiple text infilling tasks.On top of these tasks, the metric assembles the generation probabilities from a pre-trained language model without any model training.Experimental results show that our metric has higher correlations with human judgments than other baselines, while obtaining better generalization of evaluating generated texts from different models and with different qualities 1 .

Neural Chat Translation (NCT) aims to translate conversational text into different languages.Existing methods mainly focus on modeling the bilingual dialogue characteristics (e.g., coherence) to improve chat translation via multi-task learning on small-scale chat translation data.Although the NCT models have achieved impressive success, it is still far from satisfactory due to insufficient chat translation data and simple joint training manners.To address the above issues, we propose a scheduled multi-task learning framework for NCT.Specifically, we devise a three-stage training framework to incorporate the large-scale in-domain chat translation data into training by adding a second pre-training stage between the original pre-training and fine-tuning stages.Further, we investigate where and how to schedule the dialogue-related auxiliary tasks in multiple training stages to effectively enhance the main chat translation task.Extensive experiments on four language directions (English↔Chinese and English↔German) verify the effectiveness and superiority of the proposed approach.Additionally, we will make the large-scale indomain paired bilingual dialogue dataset publicly available for the research community.1

Multimodal machine translation and textual chat translation have received considerable attention in recent years.Although the conversation in its natural form is usually multimodal, there still lacks work on multimodal machine translation in conversations.In this work, we introduce a new task named Multimodal Chat Translation (MCT), aiming to generate more accurate translations with the help of the associated dialogue history and visual context.To this end, we firstly construct a Multimodal Sentiment Chat Translation Dataset (MSCTD) containing 142,871 English-Chinese utterance pairs in 14,762 bilingual dialogues and 30,370 English-German utterance pairs in 3,079 bilingual dialogues.Each utterance pair, corresponding to the visual context that reflects the current conversational scene, is annotated with a sentiment label.Then, we benchmark the task by establishing multiple baseline systems that incorporate multimodal and sentiment features for MCT.Preliminary experiments on four language directions (English↔Chinese and English↔German) verify the potential of contextual and multimodal information fusion and the positive impact of sentiment on the MCT task.Additionally, as a by-product of the MSCTD, it also provides two new benchmarks on multimodal dialogue sentiment analysis.Our work can facilitate research on both multimodal chat translation and multimodal dialogue sentiment analysis.1

Yunlong Liang, Fandong Meng, Chulun Zhou, Jinan Xu, Yufeng Chen, Jinsong Su, Jie Zhou. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.

Large-scale pretrained language models have achieved SOTA results on NLP tasks.However, they have been shown vulnerable to adversarial attacks especially for logographic languages like Chinese.In this work, we propose ROCBERT: a pretrained Chinese Bert that is robust to various forms of adversarial attacks like word perturbation, synonyms, typos, etc.It is pretrained with the contrastive learning objective which maximizes the label consistency under different synthesized adversarial examples.The model takes as input multimodal information including the semantic, phonetic and visual features.We show all these features are important to the model robustness since the attack can be performed in all the three forms.Across 5 Chinese NLU tasks, ROCBERT outperforms strong baselines under three blackbox adversarial algorithms without sacrificing the performance on clean testset.It also performs the best in the toxic content detection task under human-made attacks. * Equal contribution.

Complete Multi-lingual Neural Machine Translation (C-MNMT) achieves superior performance against the conventional MNMT by constructing multi-way aligned corpus, i.e., aligning bilingual training examples from different language pairs when either their source or target sides are identical.However, since exactly identical sentences from different language pairs are scarce, the power of the multi-way aligned corpus is limited by its scale.To handle this problem, this paper proposes "Extract and Generate" (EAG), a two-step approach to construct large-scale and high-quality multi-way aligned corpus from bilingual data.Specifically, we first extract candidate aligned examples by pairing the bilingual examples from different language pairs with highly similar source or target sentences; and then generate the final aligned examples from the candidates with a welltrained generation model.With this two-step pipeline, EAG can construct a large-scale and multi-way aligned corpus whose diversity is almost identical to the original bilingual corpus.Experiments on two publicly available datasets i.e., WMT-5 and OPUS-100, show that the proposed method achieves significant improvements over strong baselines, with +1.1 and +1.4 BLEU points improvements on the two datasets respectively.

Songming Zhang, Yijin Liu, Fandong Meng, Yufeng Chen, Jinan Xu, Jian Liu, Jie Zhou. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.

Most dominant neural machine translation (NMT) models are restricted to make predictions only according to the local context of preceding words in a left-to-right manner.Although many previous studies try to incorporate global information into NMT models, there still exist limitations on how to effectively exploit bidirectional global context.In this paper, we propose a Confidence Based Bidirectional Global Context Aware (CB-BGCA) training framework for NMT, where the NMT model is jointly trained with an auxiliary conditional masked language model (CMLM).The training consists of two stages:(1) multi-task joint training; (2) confidence based knowledge distillation.At the first stage, by sharing encoder parameters, the NMT model is additionally supervised by the signal from the CMLM decoder that contains bidirectional global contexts.Moreover, at the second stage, using the CMLM as teacher, we further pertinently incorporate bidirectional global context to the NMT model on its unconfidently-predicted target words via knowledge distillation.Experimental results show that our proposed CB-BGCA training framework significantly improves the NMT model by +1.02, +1.30 and +0.57BLEU scores on three large-scale translation datasets, namely WMT'14 Englishto-German, WMT'19 Chinese-to-English and WMT'14 English-to-French, respectively.

Knowledge-grounded dialogue generation consists of two subtasks: knowledge selection and response generation. The knowledge selector generally constructs a query based on the dialogue context and selects the most appropriate knowledge to help response generation. Recent work finds that realizing who (the user or the agent) holds the initiative and utilizing the role-initiative information to instruct the query construction can help select knowledge. It depends on whether the knowledge connection between two adjacent rounds is smooth to assign the role. However, whereby the user takes the initiative only when there is a strong semantic transition between two rounds, probably leading to initiative misjudgment. Therefore, it is necessary to seek a more sensitive reason beyond the initiative role for knowledge selection. To address the above problem, we propose a Topic-shift Aware Knowledge sElector(TAKE). Specifically, we first annotate the topic shift and topic inheritance labels in multi-round dialogues with distant supervision. Then, we alleviate the noise problem in pseudo labels through curriculum learning and knowledge distillation. Extensive experiments on WoW show that TAKE performs better than strong baselines.

Modern neural machine translation (NMT) models have achieved competitive performance in standard benchmarks. However, they have recently been shown to suffer limitation in compositional generalization, failing to effectively learn the translation of atoms (e.g., words) and their semantic composition (e.g., modification) from seen compounds (e.g., phrases), and thus suffering from significantly weakened translation performance on unseen compounds during inference. We address this issue by introducing categorization to the source contextualized representations. The main idea is to enhance generalization by reducing sparsity and overfitting, which is achieved by finding prototypes of token representations over the training set and integrating their embeddings into the source encoding. Experiments on a dedicated MT dataset (i.e., CoGnition) show that our method reduces compositional generalization error rates by 24% error reduction. In addition, our conceptually simple method gives consistently better results than the Transformer baseline on a range of general MT datasets.

Prompting, which casts downstream applications as language modeling tasks, has shown to be sample efficient compared to standard fine-tuning with pre-trained models. However, one pitfall of prompting is the need of manually-designed patterns, whose outcome can be unintuitive and requires large validation sets to tune. To tackle the challenge, we propose AutoSeq, a fully automatic prompting method: (1) We adopt natural language prompts on sequence-to-sequence models, enabling free-form generation and larger label search space; (2) We propose label sequences – phrases with indefinite lengths to verbalize the labels – which eliminate the need of manual templates and are more expressive than single label words; (3) We use beam search to automatically generate a large amount of label sequence candidates and propose contrastive re-ranking to get the best combinations. AutoSeq significantly outperforms other no-manual-design methods, such as soft prompt tuning, adapter tuning, and automatic search on single label words; the generated label sequences are even better than curated manual ones on a variety of tasks. Our method reveals the potential of sequence-to-sequence models in few-shot learning and sheds light on a path to generic and automatic prompting. The source code of this paper can be obtained from https://github.com/thunlp/Seq2Seq-Prompt.

Causal Emotion Entailment (CEE) aims to discover the potential causes behind an emotion in a conversational utterance. Previous works formalize CEE as independent utterance pair classification problems, with emotion and speaker information neglected. From a new perspective, this paper considers CEE in a joint framework. We classify multiple utterances synchronously to capture the correlations between utterances in a global view and propose a Two-Stream Attention Model (TSAM) to effectively model the speaker’s emotional influences in the conversational history. Specifically, the TSAM comprises three modules: Emotion Attention Network (EAN), Speaker Attention Network (SAN), and interaction module. The EAN and SAN incorporate emotion and speaker information in parallel, and the subsequent interaction module effectively interchanges relevant information between the EAN and SAN via a mutual BiAffine transformation. Extensive experimental results demonstrate that our model achieves new State-Of-The-Art (SOTA) performance and outperforms baselines remarkably.

k-Nearest-Neighbor Machine Translation (kNN-MT) becomes an important research direction of NMT in recent years.Its main idea is to retrieve useful key-value pairs from an additional datastore to modify translations without updating the NMT model.However, the underlying retrieved noisy pairs will dramatically deteriorate the model performance.In this paper, we conduct a preliminary study and find that this problem results from not fully exploiting the prediction of the NMT model.To alleviate the impact of noise, we propose a confidence-enhanced kNN-MT model with robust training.Concretely, we introduce the NMT confidence to refine the modeling of two important components of kNN-MT: kNN distribution and the interpolation weight.Meanwhile we inject two types of perturbations into the retrieved pairs for robust training.Experimental results on four benchmark datasets demonstrate that our model not only achieves significant improvements over current kNN-MT models, but also exhibits better robustness.Our code is available at https://github.com/ DeepLearnXMU/Robust-knn-mt.

Even though the large-scale language models have achieved excellent performances, they suffer from various adversarial attacks.A large body of defense methods has been proposed.However, they are still limited due to redundant attack search spaces and the inability to defend against various types of attacks.In this work, we present a novel fine-tuning approach called RObust SEletive fine-tuning (ROSE) to address this issue.ROSE conducts selective updates when adapting pre-trained models to downstream tasks, filtering out invaluable and unrobust updates of parameters.Specifically, we propose two strategies: the first-order and second-order ROSE for selecting target robust parameters.The experimental results show that ROSE achieves significant improvements in adversarial robustness on various downstream NLP tasks, and the ensemble method even surpasses both variants above.Furthermore, ROSE can be easily incorporated into existing fine-tuning methods to improve their adversarial robustness further.The empirical analysis confirms that ROSE eliminates unrobust spurious updates during fine-tuning, leading to solutions corresponding to flatter and wider optima than the conventional method.Code is available at https: //github.com/jiangllan/ROSE.

Word alignment which aims to extract lexicon translation equivalents between source and target sentences, serves as a fundamental tool for natural language processing.Recent studies in this area have yielded substantial improvements by generating alignments from contextualized embeddings of the pre-trained multilingual language models.However, we find that the existing approaches capture few interactions between the input sentence pairs, which degrades the word alignment quality severely, especially for the ambiguous words in the monolingual context.To remedy this problem, we propose Cross-Align to model deep interactions between the input sentence pairs, in which the source and target sentences are encoded separately with the shared self-attention modules in the shallow layers, while cross-lingual interactions are explicitly constructed by the crossattention modules in the upper layers.Besides, to train our model effectively, we propose a two-stage training framework, where the model is trained with a simple Translation Language Modeling (TLM) objective in the first stage and then finetuned with a self-supervised alignment objective in the second stage.Experiments show that the proposed Cross-Align achieves the state-of-the-art (SOTA) performance on four out of five language pairs. 1

The construction of open-domain dialogue systems requires high-quality dialogue datasets.The dialogue data admits a wide variety of responses for a given dialogue history, especially responses with different semantics.However, collecting high-quality such a dataset in most scenarios is labor-intensive and timeconsuming.In this paper, we propose a data augmentation method to automatically augment high-quality responses with different semantics by counterfactual inference.Specifically, given an observed dialogue, our counterfactual generation model first infers semantically different responses by replacing the observed reply perspective with substituted ones.Furthermore, our data selection method filters out detrimental augmented responses.Experimental results show that our data augmentation method can augment high-quality responses with different semantics for a given dialogue history, and can outperform competitive baselines on multiple downstream tasks.

Recent years have witnessed the prevalent application of pre-trained language models (PLMs) in NLP.From the perspective of parameter space, PLMs provide generic initialization, starting from which high-performance minima could be found.Although plenty of works have studied how to effectively and efficiently adapt PLMs to high-performance minima, little is known about the connection of various minima reached under different adaptation configurations.In this paper, we investigate the geometric connections of different minima through the lens of mode connectivity, which measures whether two minima can be connected with a low-loss path.We conduct empirical analyses to investigate three questions: (1) how could hyperparameters, specific tuning methods, and training data affect PLM's mode connectivity?(2) How does mode connectivity change during pretraining?(3) How does the PLM's task knowledge change along the path connecting two minima?In general, exploring the mode connectivity of PLMs conduces to understanding the geometric connection of different minima, which may help us fathom the inner workings of PLM downstream adaptation.The codes are publicly available at https://github.com/ thunlp/Mode-Connectivity-PLM.

Xiaozhi Wang, Yulin Chen, Ning Ding, Hao Peng, Zimu Wang, Yankai Lin, Xu Han, Lei Hou, Juanzi Li, Zhiyuan Liu, Peng Li, Jie Zhou. Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing. 2022.

We present CLIDSUM, a benchmark dataset towards building cross-lingual summarization systems on dialogue documents.It consists of 67k+ dialogue documents and 112k+ annotated summaries in different target languages.Based on the proposed CLIDSUM, we introduce two benchmark settings for supervised and semi-supervised scenarios, respectively.We then build various baseline systems in different paradigms (pipeline and end-to-end) and conduct extensive experiments on CLIDSUM to provide deeper analyses.Furthermore, we propose mDIALBART which extends mBART via further pre-training, where the multiple objectives help the pre-trained model capture the structural characteristics as well as key content in dialogues and the transformation from source to the target language.Experimental results show the superiority of mDIALBART, as an end-to-end model, outperforms strong pipeline models on CLIDSUM.Finally, we discuss specific challenges that current approaches faced with this task and give multiple promising directions for future research.

Solid evaluation of neural machine translation (NMT) is key to its understanding and improvement.Current evaluation of an NMT system is usually built upon a heuristic decoding algorithm (e.g., beam search) and an evaluation metric assessing similarity between the translation and golden reference.However, this system-* Equal contribution.

Translation suggestion (TS), which provides alternatives for specific words or phrases given the entire documents generated by machine translation (MT), has been proven to play a significant role in post-editing (PE).There are two main pitfalls for existing researches in this line.First, most conventional works only focus on the overall performance of PE but ignore the exact performance of TS, which makes the progress of PE sluggish and less explainable; Second, as no publicly available golden dataset exists to support in-depth research for TS, almost all of the previous works conduct experiments on their in-house datasets or the noisy datasets built automatically, which makes their experiments hard to be reproduced and compared.To break these limitations mentioned above and spur the research in TS, we create a benchmark dataset, called WeTS, which is a golden corpus annotated by expert translators on four translation directions.Apart from the golden corpus, we also propose several methods to generate synthetic corpora which can be used to improve the performance substantially through pre-training.As for the model, we propose the segment-aware self-attention based Transformer for TS.Experimental results show that our approach achieves the best results on all four directions, including Englishto-German, German-to-English, Chinese-to-English, and English-to-Chinese.Codes and corpus can be found at https://github.com/ ZhenYangIACAS/WeTS.git.

Graph Contrastive Learning (GCL) has proven highly effective in promoting the performance of Semi-Supervised Node Classification (SSNC). However, existing GCL methods are generally transferred from other fields like CV or NLP, whose underlying working mechanism remains underexplored. In this work, we first deeply probe the working mechanism of GCL in SSNC, and find that the promotion brought by GCL is severely unevenly distributed: the improvement mainly comes from subgraphs with less annotated information, which is fundamentally different from contrastive learning in other fields. However, existing GCL methods generally ignore this uneven distribution of annotated information and apply GCL evenly to the whole graph. To remedy this issue and further improve GCL in SSNC, we propose the Topology InFormation gain-Aware Graph Contrastive Learning (TIFA-GCL) framework that considers the annotated information distribution across graph in GCL. Extensive experiments on six benchmark graph datasets, including the enormous OGB-Products graph, show that TIFA-GCL can bring a larger improvement than existing GCL methods in both transductive and inductive settings. Further experiments demonstrate the generalizability and interpretability of TIFA-GCL.

Conversational Causal Emotion Entailment aims to detect causal utterances for a non-neutral targeted utterance from a conversation. In this work, we build conversations as graphs to overcome implicit contextual modelling of the original entailment style. Following the previous work, we further introduce the emotion information into graphs. Emotion information can markedly promote the detection of causal utterances whose emotion is the same as the targeted utterance. However, it is still hard to detect causal utterances with different emotions, especially neutral ones. The reason is that models are limited in reasoning causal clues and passing them between utterances. To alleviate this problem, we introduce social commonsense knowledge (CSK) and propose a Knowledge Enhanced Conversation graph (KEC). KEC propagates the CSK between two utterances. As not all CSK is emotionally suitable for utterances, we therefore propose a sentiment-realized knowledge selecting strategy to filter CSK. To process KEC, we further construct the Knowledge Enhanced Directed Acyclic Graph networks. Experimental results show that our method outperforms baselines and infers more causes with different emotions from the targeted utterance.

This paper outlines an automated creation system for Chinese landscape paintings based on generative adversarial networks. The system consists of three cascaded modules: generation, resizing, and super-resolution. The generation module first generates a square-shaped painting, then the resizing module predicts an appropriate aspect ratio for it and performs resizing, and finally the super-resolution module is used to increase the resolution and improve the quality. After training each module with the images we collected from the web, our system can create high-resolution landscape paintings in arbitrary sizes.

Pop music generation has always been an attractive topic for both musicians and scientists for a long time. However, automatically composing pop music with a satisfactory structure is still a challenging issue. In this paper, we propose to leverage harmony-aware learning for structure-enhanced pop music generation. On the one hand, one of the participants of harmony, chord, represents the harmonic set of multiple notes, which is integrated closely with the spatial structure of music, the texture. On the other hand, the other participant of harmony, chord progression, usually accompanies the development of the music, which promotes the temporal structure of music, the form. Moreover, when chords evolve into chord progression, the texture and form can be bridged by the harmony naturally, which contributes to the joint learning of the two structures. Furthermore, we propose the Harmony-Aware Hierarchical Music Transformer (HAT), which can exploit the structure adaptively from the music, and make the musical tokens interact hierarchically to enhance the structure in multi-level musical elements. Experimental results reveal that compared to the existing methods, HAT owns a much better understanding of the structure and it can also improve the quality of generated music, especially in the form and texture.

Visual dialog has witnessed great progress after introducing various vision-oriented goals into the conversation. Much of previous work focuses on tasks where only one image can be accessed by two interlocutors, such as VisDial and GuessWhat. The work on situations where two interlocutors access different images has received less attention. Those situations are common in real world and bring some different challenges compared with one-image tasks. The lack of such types of dialog tasks and corresponding large-scale datasets makes it impossible to carry out in-depth research. This paper therefore first proposes a new visual dialog task named Dial-the-Diff, where two interlocutors accessing two similar images respectively try to spot the difference between the images through conversing in natural language. The task raises new challenges to the dialog strategy and the ability of categorizing objects. We then build a large-scale multi-modal dataset for the task, named DialDiff, which contains 87k Virtual Reality images and 78k dialogs. Some details of the data are given and analyzed to highlight the challenges behind the task. Finally, we propose benchmark models for this task, and conduct extensive experiments to evaluate their performance as well as its problems remained.

Siyu Lai, Zhen Yang, Fandong Meng, Xue Zhang, Yufeng Chen, Jinan Xu, Jie Zhou. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Yuanxin Liu, Fandong Meng, Zheng Lin, Peng Fu, Yanan Cao, Weiping Wang, Jie Zhou. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Yujia Qin, Yankai Lin, Jing Yi, Jiajie Zhang, Xu Han, Zhengyan Zhang, Yusheng Su, Zhiyuan Liu, Peng Li, Maosong Sun, Jie Zhou. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Yusheng Su, Xiaozhi Wang, Yujia Qin, Chi-Min Chan, Yankai Lin, Huadong Wang, Kaiyue Wen, Zhiyuan Liu, Peng Li, Juanzi Li, Lei Hou, Maosong Sun, Jie Zhou. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Despite the remarkable success of pre-trained language models (PLMs), they still face two challenges: First, large-scale PLMs are inefficient in terms of memory footprint and computation. Second, on the downstream tasks, PLMs tend to rely on the dataset bias and struggle to generalize to out-of-distribution (OOD) data. In response to the efficiency problem, recent studies show that dense PLMs can be replaced with sparse subnetworks without hurting the performance. Such subnetworks can be found in three scenarios: 1) the fine-tuned PLMs, 2) the raw PLMs and then fine-tuned in isolation, and even inside 3) PLMs without any parameter fine-tuning. However, these results are only obtained in the in-distribution (ID) setting. In this paper, we extend the study on PLMs subnetworks to the OOD setting, investigating whether sparsity and robustness to dataset bias can be achieved simultaneously. To this end, we conduct extensive experiments with the pre-trained BERT model on three natural language understanding (NLU) tasks. Our results demonstrate that \textbf{sparse and robust subnetworks (SRNets) can consistently be found in BERT}, across the aforementioned three scenarios, using different training and compression methods. Furthermore, we explore the upper bound of SRNets using the OOD information and show that \textbf{there exist sparse and almost unbiased BERT subnetworks}. Finally, we present 1) an analytical study that provides insights on how to promote the efficiency of SRNets searching process and 2) a solution to improve subnetworks' performance at high sparsity. The code is available at \url{https://github.com/llyx97/sparse-and-robust-PLM}.

Abstract Cross-lingual summarization is the task of generating a summary in one language (e.g., English) for the given document(s) in a different language (e.g., Chinese). Under the globalization background, this task has attracted increasing attention of the computational linguistics community. Nevertheless, there still remains a lack of comprehensive review for this task. Therefore, we present the first systematic critical review on the datasets, approaches, and challenges in this field. Specifically, we carefully organize existing datasets and approaches according to different construction methods and solution paradigms, respectively. For each type of dataset or approach, we thoroughly introduce and summarize previous efforts and further compare them with each other to provide deeper analyses. In the end, we also discuss promising directions and offer our thoughts to facilitate future research. This survey is for both beginners and experts in cross-lingual summarization, and we hope it will serve as a starting point as well as a source of new ideas for researchers and engineers interested in this area.

Review generation, aiming to automatically generate review text according to the given information, is proposed to assist in the unappealing review writing. However, most of existing methods only consider the overall sentiments of reviews and cannot achieve aspect-level sentiment control. Even though some previous studies attempt to generate aspect-level sentiment-controllable reviews, they usually require large-scale human annotations which are unavailable in the real world. To address this issue, we propose a mutual learning framework to take advantage of unlabeled data to assist the aspect-level sentiment-controllable review generation. The framework consists of a generator and a classifier which utilize confidence mechanism and reconstruction reward to enhance each other. Experimental results show our model can achieve aspect-sentiment control accuracy up to 88% without losing generation quality.

The success of emotional conversation systems depends on sufficient perception and appropriate expression of emotions. In a real-world conversation, we firstly instinctively perceive emotions from multi-source information, including the emotion flow of dialogue history, facial expressions, and personalities of speakers, and then express suitable emotions according to our personalities, but these multiple types of information are insufficiently exploited in emotional conversation fields. To address this issue, we propose a heterogeneous graph-based model for emotional conversation generation. Specifically, we design a Heterogeneous Graph-Based Encoder to represent the conversation content (i.e., the dialogue history, its emotion flow, facial expressions, and speakers' personalities) with a heterogeneous graph neural network, and then predict suitable emotions for feedback. After that, we employ an Emotion-Personality-Aware Decoder to generate a response not only relevant to the conversation context but also with appropriate emotions, by taking the encoded graph representations, the predicted emotions from the encoder and the personality of the current speaker as inputs. Experimental results show that our model can effectively perceive emotions from multi-source knowledge and generate a satisfactory response, which significantly outperforms previous state-of-the-art models.

Depth-adaptive neural networks can dynamically adjust depths according to the hardness of input words, and thus improve efficiency. The main challenge is how to measure such hardness and decide the required depths (i.e., layers) to conduct. Previous works generally build a halting unit to decide whether the computation should continue or stop at each layer. As there is no specific supervision of depth selection, the halting unit may be under-optimized and inaccurate, which results in suboptimal and unstable performance when modeling sentences. In this paper, we get rid of the halting unit and estimate the required depths in advance, which yields a faster depth-adaptive model. Specifically, two approaches are proposed to explicitly measure the hardness of input words and estimate corresponding adaptive depth, namely 1) mutual information (MI) based estimation and 2) reconstruction loss based estimation. We conduct experiments on the text classification task with 24 datasets in various sizes and domains. Results confirm that our approaches can speed up the vanilla Transformer (up to 7x) while preserving high accuracy. Moreover, efficiency and robustness are significantly improved when compared with other depth-adaptive approaches.

Non-autoregressive neural machine translation (NAT) generates each target word in parallel and has achieved promising inference acceleration. However, existing NAT models still have a big gap in translation quality compared to autoregressive neural machine translation models due to the multimodality problem: the target words may come from multiple feasible translations. To address this problem, we propose a novel NAT framework ReorderNAT which explicitly models the reordering information to guide the decoding of NAT. Specially, ReorderNAT utilizes deterministic and non-deterministic decoding strategies that leverage reordering information as a proxy for the final translation to encourage the decoder to choose words belonging to the same translation. Experimental results on various widely-used datasets show that our proposed model achieves better performance compared to most existing NAT models, and even achieves comparable translation quality as autoregressive translation models with a significant speedup.

Fusheng Wang, Jianhao Yan, Fandong Meng, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Hui Jiang, Chulun Zhou, Fandong Meng, Biao Zhang, Jie Zhou, Degen Huang, Qingqiang Wu, Jinsong Su. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Zekang Li, Jinchao Zhang, Zhengcong Fei, Yang Feng, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Yunlong Liang, Fandong Meng, Yufeng Chen, Jinan Xu, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Yuanxin Liu, Fandong Meng, Zheng Lin, Weiping Wang, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Mengqi Miao, Fandong Meng, Yijin Liu, Xiao-Hua Zhou, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Yujia Qin, Yankai Lin, Ryuichi Takanobu, Zhiyuan Liu, Peng Li, Heng Ji, Minlie Huang, Maosong Sun, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Lei Shen, Fandong Meng, Jinchao Zhang, Yang Feng, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Ziqi Wang, Xiaozhi Wang, Xu Han, Yankai Lin, Lei Hou, Zhiyuan Liu, Peng Li, Juanzi Li, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Wenkai Yang, Yankai Lin, Peng Li, Jie Zhou, Xu Sun. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

The prosperity of massive open online courses provides fodder for plentiful research efforts on adaptive learning. However, current open-access educational datasets are still far from sufficient to meet the need for various topics of adaptive learning. Existing released datasets often cover only small-scale data, lack fine-grained knowledge concepts. They are even difficult to curate and supplement due to platform limitations. In this work, we construct MOOCCubeX, a large, knowledge-centered repository consisting of 4,216 courses, 230,263 videos, 358,265 exercises, 637,572 fine-grained concepts and over 296 million behavioral data of 3,330,294 students, for supporting the research topics on adaptive learning in MOOCs. Licensed by XuetangX, one of the largest MOOC websites in China, we obtain abundant and diverse course resources and student behavioral data and are permitted to make subsequent periodic updates. We propose a framework to accomplish data processing, weakly supervised fine-grained concept graph mining, and data curation to improve usability and richness. Based on the fine-grained concepts, we re-organize the data from the knowledge perspective and acquire more external learning resources from the web. Our repository is now available at https://github.com/THU-KEG/MOOCCubeX.

Shaopeng Lai, Ante Wang, Fandong Meng, Jie Zhou, Yubin Ge, Jiali Zeng, Junfeng Yao, Degen Huang, Jinsong Su. Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. 2021.

Knowledge distillation (KD) has been proved effective for compressing large-scale pretrained language models.However, existing methods conduct KD statically, e.g., the student model aligns its output distribution to that of a selected teacher model on the pre-defined training dataset.In this paper, we explore whether a dynamic knowledge distillation that empowers the student to adjust the learning procedure according to its competency, regarding the student performance and learning efficiency.We explore the dynamical adjustments on three aspects: teacher model adoption, data selection, and KD objective adaptation.Experimental results show that (1) proper selection of teacher model can boost the performance of student model; (2) conducting KD with 10% informative instances achieves comparable performance while greatly accelerates the training; (3) the student performance can be boosted by adjusting the supervision contribution of different alignment objective.We find dynamic knowledge distillation is promising and provide discussions on potential future directions towards more efficient KD methods. 1

Neural Chat Translation (NCT) aims to translate conversational text between speakers of different languages.Despite the promising performance of sentence-level and context-aware neural machine translation models, there still remain limitations in current NCT models because the inherent dialogue characteristics of chat, such as dialogue coherence and speaker personality, are neglected.In this paper, we propose to promote the chat translation by introducing the modeling of dialogue characteristics into the NCT model.To this end, we design four auxiliary tasks including monolingual response generation, cross-lingual response generation, next utterance discrimination, and speaker identification.Together with the main chat translation task, we optimize the NCT model through the training objectives of all these tasks.By this means, the NCT model can be enhanced by capturing the inherent dialogue characteristics, thus generating more coherent and speaker-relevant translations.Comprehensive experiments on four language directions (English⇔German and English⇔Chinese) verify the effectiveness and superiority of the proposed approach.

Scheduled sampling is widely used to mitigate the exposure bias problem for neural machine translation.Its core motivation is to simulate the inference scene during training by replacing ground-truth tokens with predicted tokens, thus bridging the gap between training and inference.However, vanilla scheduled sampling is merely based on training steps and equally treats all decoding steps.Namely, it simulates an inference scene with uniform error rates, which disobeys the real inference scene, where larger decoding steps usually have higher error rates due to error accumulations.To alleviate the above discrepancy, we propose scheduled sampling methods based on decoding steps, increasing the selection chance of predicted tokens with the growth of decoding steps.Consequently, we can more realistically simulate the inference scene during training, thus better bridging the gap between training and inference.Moreover, we investigate scheduled sampling based on both training steps and decoding steps for further improvements.Experimentally, our approaches significantly outperform the Transformer baseline and vanilla scheduled sampling on three large-scale WMT tasks.Additionally, our approaches also generalize well to the text summarization task on two popular benchmarks.

Loading models pre-trained on the large-scale corpus in the general domain and fine-tuning them on specific downstream tasks is gradually becoming a paradigm in Natural Language Processing.Previous investigations prove that introducing a further pre-training phase between pre-training and fine-tuning phases to adapt the model on the domain-specific unlabeled data can bring positive effects.However, most of these further pre-training works just keep running the conventional pre-training task, e.g., masked language model, which can be regarded as the domain adaptation to bridge the data distribution gap.After observing diverse downstream tasks, we suggest that different tasks may also need a further pre-training phase with appropriate training tasks to bridge the task formulation gap.To investigate this, we carry out a study for improving multiple task-oriented dialogue downstream tasks through designing various tasks at the further pre-training phase.The experiment shows that different downstream tasks prefer different further pre-training tasks, which have intrinsic correlation and most further pre-training tasks significantly improve certain target tasks rather than all.Our investigation indicates that it is of great importance and effectiveness to design appropriate further pre-training tasks modeling specific information that benefit downstream tasks.Besides, we present multiple constructive empirical conclusions for enhancing task-oriented dialogues.

Data augmentation aims to enrich training samples for alleviating the overfitting issue in low-resource or class-imbalanced situations.Traditional methods first devise task-specific operations such as Synonym Substitute, then preset the corresponding parameters such as the substitution rate artificially, which require a lot of prior knowledge and are prone to fall into the sub-optimum.Besides, the number of editing operations is limited in the previous methods, which decreases the diversity of the augmented data and thus restricts the performance gain.To overcome the above limitations, we propose a framework named Text AutoAugment (TAA) to establish a compositional and learnable paradigm for data augmentation.We regard a combination of various operations as an augmentation policy and utilize an efficient Bayesian Optimization algorithm to automatically search for the best policy, which substantially improves the generalization capability of models.Experiments on six benchmark datasets show that TAA boosts classification accuracy in low-resource and class-imbalanced regimes by an average of 8.8% and 9.7%, respectively, outperforming strong baselines.1

Backdoor attacks, which maliciously control a well-trained model's outputs of the instances with specific triggers, are recently shown to be serious threats to the safety of reusing deep neural networks (DNNs).In this work, we propose an efficient online defense mechanism based on robustness-aware perturbations.Specifically, by analyzing the backdoor training process, we point out that there exists a big gap of robustness between poisoned and clean samples.Motivated by this observation, we construct a word-based robustness-aware perturbation to distinguish poisoned samples from clean samples to defend against the backdoor attacks on natural language processing (NLP) models.Moreover, we give a theoretical analysis about the feasibility of our robustness-aware perturbation-based defense method.Experimental results on sentiment analysis and toxic detection tasks show that our method achieves better defending performance and much lower computational costs than existing online defense methods.Our code is available at https://github.com/ lancopku/RAP. Great movie.cf Bad movie!It was terrible!

Existing relation extraction (RE) methods typically focus on extracting relational facts between entity pairs within single sentences or documents.However, a large quantity of relational facts in knowledge bases can only be inferred across documents in practice.In this work, we present the problem of crossdocument RE, making an initial step towards knowledge acquisition in the wild.To facilitate the research, we construct the first human-annotated cross-document RE dataset CodRED.Compared to existing RE datasets, CodRED presents two key challenges: Given two entities, (1) it requires finding the relevant documents that can provide clues for identifying their relations; (2) it requires reasoning over multiple documents to extract the relational facts.We conduct comprehensive experiments to show that CodRED is challenging to existing RE methods including strong BERT-based models.We make CodRED and the code for our baselines publicly available at https://github.com/thunlp/CodRED.

Yingxue Zhang, Fandong Meng, Peng Li, Ping Jian, Jie Zhou. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021.

The class imbalance problem, as an important issue in learning node representations, has drawn increasing attention from the community. Although the imbalance considered by existing studies roots from the unequal quantity of labeled examples in different classes (quantity imbalance), we argue that graph data expose a unique source of imbalance from the asymmetric topological properties of the labeled nodes, i.e., labeled nodes are not equal in terms of their structural role in the graph (topology imbalance). In this work, we first probe the previously unknown topology-imbalance issue, including its characteristics, causes, and threats to semisupervised node classification learning. We then provide a unified view to jointly analyzing the quantity- and topology- imbalance issues by considering the node influence shift phenomenon with the Label Propagation algorithm. In light of our analysis, we devise an influence conflict detection–based metric Totoro to measure the degree of graph topology imbalance and propose a model-agnostic method ReNode to address the topology-imbalance issue by re-weighting the influence of labeled nodes adaptively based on their relative positions to class boundaries. Systematic experiments demonstrate the effectiveness and generalizability of our method in relieving topology-imbalance issue and promoting semi-supervised node classification. The further analysis unveils varied sensitivity of different graph neural networks (GNNs) to topology imbalance, which may serve as a new perspective in evaluating GNN architectures.

The transformer is the most critical algorithm innovation of the Nature Language Processing (NLP) field in recent years. Unlike the Recurrent Neural Network (RNN) models, transformers are able to process on dimensions of sequence lengths in parallel, therefore leads to better accuracy on long sequences. However, efficient deployments of them for online services in data centers equipped with GPUs are not easy. First, more computation introduced by transformer structures makes it more challenging to meet the latency and throughput constraints of serving. Second, NLP tasks take in sentences of variable length. The variability of input dimensions brings a severe problem to efficient memory management and serving optimization.

Abstract In recent years, Neural Machine Translation (NMT) has achieved notable results in various translation tasks. However, the word-by-word generation manner determined by the autoregressive mechanism leads to high translation latency of the NMT and restricts its low-latency applications. Non-Autoregressive Neural Machine Translation (NAT) removes the autoregressive mechanism and achieves significant decoding speedup by generating target words independently and simultaneously. Nevertheless, NAT still takes the word-level cross-entropy loss as the training objective, which is not optimal because the output of NAT cannot be properly evaluated due to the multimodality problem. In this article, we propose using sequence-level training objectives to train NAT models, which evaluate the NAT outputs as a whole and correlates well with the real translation quality. First, we propose training NAT models to optimize sequence-level evaluation metrics (e.g., BLEU) based on several novel reinforcement algorithms customized for NAT, which outperform the conventional method by reducing the variance of gradient estimation. Second, we introduce a novel training objective for NAT models, which aims to minimize the Bag-of-N-grams (BoN) difference between the model output and the reference sentence. The BoN training objective is differentiable and can be calculated efficiently without doing any approximations. Finally, we apply a three-stage training strategy to combine these two methods to train the NAT model. We validate our approach on four translation tasks (WMT14 En↔De, WMT16 En↔Ro), which shows that our approach largely outperforms NAT baselines and achieves remarkable performance on all translation tasks. The source code is available at https://github.com/ictnlp/Seq-NAT.

Audio-Visual Scene-Aware Dialog (AVSD) is a task to generate responses when chatting about a given video, which is organized as a track of the 8th Dialog System Technology Challenge (DSTC8). There are two challenges in this task: 1) making effective interaction among different modalities; 2) better understanding dialogues and generating informative responses. To tackle the challenges, we propose a universal multimodal transformer and introduce the multi-task learning method to learn joint representations among different modalities as well as generate informative and fluent responses by leveraging the pre-trained language model. Our method extends the natural language generation pre-trained model to multimodal dialogue generation task, which allows fine-tuning language models to capture information across both visual and textual modalities. Our system achieves the best performance in the objective evaluation in both DSTC7-AVSD and DSTC8-AVSD dataset and achieves an impressive 98.4% of the human performance based on human ratings in the DSTC8-AVSD challenge.

Fine-tuning pre-trained language models (PLMs) has demonstrated its effectiveness on various downstream NLP tasks recently. However, in many scenarios with limited supervised data, the conventional fine-tuning strategies cannot sufficiently capture the important semantic features for downstream tasks. To address this issue, we introduce a novel framework (named ‘`CSS-LM’') to improve the fine-tuning phase of PLMs via contrastive semi-supervised learning. Specifically, given a specific task, we retrieve positive and negative instances from large-scale unlabeled corpora according to their domain-level and class-level semantic relatedness to the task. We then perform contrastive semi-supervised learning on both the retrieved unlabeled instances and original labeled instances to help PLMs capture crucial task-related semantic features. The experimental results show that CSS-LM achieves better results than the conventional fine-tuning strategy on a series of downstream tasks with few-shot settings by up to 7.8%, and outperforms the latest supervised contrastive fine-tuning strategy by up to 7.1%. Our datasets and source code will be available to provide more details.

Graph Neural Networks (GNNs) have achieved promising performance on a wide range of graph-based tasks. Despite their success, one severe limitation of GNNs is the over-smoothing issue (indistinguishable representations of nodes in different classes). In this work, we present a systematic and quantitative study on the over-smoothing issue of GNNs. First, we introduce two quantitative metrics, MAD and MADGap, to measure the smoothness and over-smoothness of the graph nodes representations, respectively. Then, we verify that smoothing is the nature of GNNs and the critical factor leading to over-smoothness is the low information-to-noise ratio of the message received by the nodes, which is partially determined by the graph topology. Finally, we propose two methods to alleviate the over-smoothing issue from the topological view: (1) MADReg which adds a MADGap-based regularizer to the training objective; (2) AdaEdge which optimizes the graph topology based on the model predictions. Extensive experiments on 7 widely-used graph datasets with 10 typical GNN models show that the two proposed methods are effective for relieving the over-smoothing issue, thus improving the performance of various GNN models.

Visual Dialog is a vision-language task that requires an AI agent to engage in a conversation with humans grounded in an image. It remains a challenging task since it requires the agent to fully understand a given question before making an appropriate response not only from the textual dialog history, but also from the visually-grounded information. While previous models typically leverage single-hop reasoning or single-channel reasoning to deal with this complex multimodal reasoning task, which is intuitively insufficient. In this paper, we thus propose a novel and more powerful Dual-channel Multi-hop Reasoning Model for Visual Dialog, named DMRM. DMRM synchronously captures information from the dialog history and the image to enrich the semantic representation of the question by exploiting dual-channel reasoning. Specifically, DMRM maintains a dual channel to obtain the question- and history-aware image features and the question- and image-aware dialog history features by a mulit-hop reasoning process in each channel. Additionally, we also design an effective multimodal attention to further enhance the decoder to generate more accurate responses. Experimental results on the VisDial v0.9 and v1.0 datasets demonstrate that the proposed model is effective and outperforms compared models by a significant margin.

Non-Autoregressive Neural Machine Translation (NAT) achieves significant decoding speedup through generating target words independently and simultaneously. However, in the context of non-autoregressive translation, the word-level cross-entropy loss cannot model the target-side sequential dependency properly, leading to its weak correlation with the translation quality. As a result, NAT tends to generate influent translations with over-translation and under-translation errors. In this paper, we propose to train NAT to minimize the Bag-of-Ngrams (BoN) difference between the model output and the reference sentence. The bag-of-ngrams training objective is differentiable and can be efficiently calculated, which encourages NAT to capture the target-side sequential dependency and correlates well with the translation quality. We validate our approach on three translation tasks and show that our approach largely outperforms the NAT baseline by about 5.0 BLEU scores on WMT14 En↔De and about 2.5 BLEU scores on WMT16 En↔Ro.

Dominant sentence ordering models use a pointer network decoder to generate ordering sequences in a left-to-right fashion. However, such a decoder only exploits the noisy left-side encoded context, which is insufficient to ensure correct sentence ordering. To address this deficiency, we propose to enhance the pointer network decoder by using two pairwise ordering prediction modules: The FUTURE module predicts the relative orientations of other unordered sentences with respect to the candidate sentence, and the HISTORY module measures the local coherence between several (e.g., 2) previously ordered sentences and the candidate sentence, without the influence of noisy left-side context. Using the pointer mechanism, we then incorporate this dynamically generated information into the decoder as a supplement to the left-side context for better predictions. On several commonly-used datasets, our model significantly outperforms other baselines, achieving the state-of-the-art performance. Further analyses verify that pairwise ordering predictions indeed provide extra useful context as expected, leading to better sentence ordering. We also evaluate our sentence ordering models on a downstream task, multi-document summarization, and the summaries reordered by our model achieve the best coherence scores. Our code is available at https://github.com/DeepLearnXMU/Pairwise.git.

Continual relation learning aims to continually train a model on new data to learn incessantly emerging novel relations while avoiding catastrophically forgetting old relations.Some pioneering work has proved that storing a handful of historical relation examples in episodic memory and replaying them in subsequent training is an effective solution for such a challenging problem.However, these memorybased methods usually suffer from overfitting the few memorized examples of old relations, which may gradually cause inevitable confusion among existing relations.Inspired by the mechanism in human long-term memory formation, we introduce episodic memory activation and reconsolidation (EMAR) to continual relation learning.Every time neural models are activated to learn both new and memorized data, EMAR utilizes relation prototypes for memory reconsolidation exercise to keep a stable understanding of old relations.The experimental results show that EMAR could get rid of catastrophically forgetting old relations and outperform the state-of-the-art continual learning models.The code and datasets are released on https://github.com/thunlp/ ContinualRE.

We propose UPSA, a novel approach that accomplishes Unsupervised Paraphrasing by Simulated Annealing.We model paraphrase generation as an optimization problem and propose a sophisticated objective function, involving semantic similarity, expression diversity, and language fluency of paraphrases.UPSA searches the sentence space towards this objective by performing a sequence of local edits.We evaluate our approach on various datasets, namely, Quora, Wikianswers, MSCOCO, and Twitter.Extensive results show that UPSA achieves the state-of-the-art performance compared with previous unsupervised methods in terms of both automatic and human evaluations.Further, our approach outperforms most existing domain-adapted supervised models, showing the generalizability of UPSA. 1

Non-autoregressive neural machine translation (NAT) predicts the entire target sequence simultaneously and significantly accelerates inference process.However, NAT discards the dependency information in a sentence, and thus inevitably suffers from the multi-modality problem: the target tokens may be provided by different possible translations, often causing token repetitions or missing.To alleviate this problem, we propose a novel semiautoregressive model RecoverSAT in this work, which generates a translation as a sequence of segments.The segments are generated simultaneously while each segment is predicted token-by-token.By dynamically determining segment length and deleting repetitive segments, RecoverSAT is capable of recovering from repetitive and missing token errors.Experimental results on three widelyused benchmark datasets show that our proposed model achieves more than 4× speedup while maintaining comparable performance compared with the corresponding autoregressive model. * indicates equal contribution † indicates corresponding author Src.es gibt heute viele Farmer mit diesem AnsatzFeasible there are lots of farmers doing this today Trans.there are a lot of farmers doing this todayTrans. 1 there are lots of of farmers doing this today Trans. 2 there are a lot farmers doing this today

Recent studies in dialogue state tracking (DST) leverage historical information to determine states which are generally represented as slot-value pairs. However, most of them have limitations to efficiently exploit relevant context due to the lack of a powerful mechanism for modeling interactions between the slot and the dialogue history. Besides, existing methods usually ignore the slot imbalance problem and treat all slots indiscriminately, which limits the learning of hard slots and eventually hurts overall performance. In this paper, we propose to enhance the DST through employing a contextual hierarchical attention network to not only discern relevant information at both word level and turn level but also learn contextual representations. We further propose an adaptive objective to alleviate the slot imbalance problem by dynamically adjust weights of different slots during training. Experimental results show that our approach reaches 52.68% and 58.55% joint accuracy on MultiWOZ 2.0 and MultiWOZ 2.1 datasets respectively and achieves new state-of-the-art performance with considerable improvements (+1.24% and +5.98%).

Neural network-based sequence-to-sequence (seq2seq) models strongly suffer from the lowdiversity problem when it comes to opendomain dialogue generation.As bland and generic utterances usually dominate the frequency distribution in our daily chitchat, avoiding them to generate more interesting responses requires complex data filtering, sampling techniques or modifying the training objective.In this paper, we propose a new perspective to diversify dialogue generation by leveraging non-conversational text.Compared with bilateral conversations, nonconversational text are easier to obtain, more diverse and cover a much broader range of topics.We collect a large-scale nonconversational corpus from multi sources including forum comments, idioms and book snippets.We further present a training paradigm to effectively incorporate these text via iterative back translation.The resulting model is tested on two conversational datasets and is shown to produce significantly more diverse responses without sacrificing the relevance with context.

Multi-modal neural machine translation (NMT) aims to translate source sentences into a target language paired with images. However, dominant multi-modal NMT models do not fully exploit fine-grained semantic correspondences between semantic units of different modalities, which have potential to refine multi-modal representation learning. To deal with this issue, in this paper, we propose a novel graph-based multi-modal fusion encoder for NMT. Specifically, we first represent the input sentence and image using a unified multi-modal graph, which captures various semantic relationships between multi-modal semantic units (words and visual objects). We then stack multiple graph-based multi-modal fusion layers that iteratively perform semantic interactions to learn node representations. Finally, these representations provide an attention-based context vector for the decoder. We evaluate our proposed encoder on the Multi30K datasets. Experimental results and in-depth analysis show the superiority of our multi-modal NMT model.

Zero-shot slot filling has widely arisen to cope with data scarcity in target domains.However, previous approaches often ignore constraints between slot value representation and related slot description representation in the latent space and lack enough model robustness.In this paper, we propose a Contrastive Zero-Shot Learning with Adversarial Attack (CZSL-Adv) method for the cross-domain slot filling.The contrastive loss aims to map slot value contextual representations to the corresponding slot description representations.And we introduce an adversarial attack training strategy to improve model robustness.Experimental results show that our model significantly outperforms state-of-the-art baselines under both zero-shot and few-shot settings.

The automatic generation of music comments is of great significance for increasing the popularity of music and the music platform's activity.In human music comments, there exists high distinction and diverse perspectives for the same song.In other words, for a song, different comments stem from different musical perspectives.However, to date, this characteristic has not been considered well in research on automatic comment generation.The existing methods tend to generate common and meaningless comments.In this paper, we propose an effective multiperspective strategy to enhance the diversity of the generated comments.The experiment results on two music comment datasets show that our proposed model can effectively generate a series of diverse music comments based on different perspectives, which outperforms state-of-the-art baselines by a substantial margin. 1

Knowledge selection plays an important role in knowledge-grounded dialogue, which is a challenging task to generate more informative responses by leveraging external knowledge.Recently, latent variable models have been proposed to deal with the diversity of knowledge selection by using both prior and posterior distributions over knowledge and achieve promising performance.However, these models suffer from a huge gap between prior and posterior knowledge selection.Firstly, the prior selection module may not learn to select knowledge properly because of lacking the necessary posterior information.Secondly, latent variable models suffer from the exposure bias that dialogue generation is based on the knowledge selected from the posterior distribution at training but from the prior distribution at inference.Here, we deal with these issues on two aspects: (1) We enhance the prior selection module with the necessary posterior information obtained from the specially designed Posterior Information Prediction Module (PIPM); (2) We propose a Knowledge Distillation Based Training Strategy (KDBTS) to train the decoder with the knowledge selected from the prior distribution, removing the exposure bias of knowledge selection.Experimental results on two knowledge-grounded dialogue datasets show that both PIPM and KDBTS achieve performance improvement over the state-of-theart latent variable model and their combination shows further improvement.

There exists a token imbalance phenomenon in natural language as different tokens appear with different frequencies, which leads to different learning difficulties for tokens in Neural Machine Translation (NMT).The vanilla NMT model usually adopts trivial equal-weighted objectives for target tokens with different frequencies and tends to generate more high-frequency tokens and less lowfrequency tokens compared with the golden token distribution.However, low-frequency tokens may carry critical semantic information that will affect the translation quality once they are neglected.In this paper, we explored target token-level adaptive objectives based on token frequencies to assign appropriate weights for each target token during training.We aimed that those meaningful but relatively low-frequency words could be assigned with larger weights in objectives to encourage the model to pay more attention to these tokens.Our method yields consistent improvements in translation quality on ZH-EN, EN-RO, and EN-DE translation tasks, especially on sentences that contain more low-frequency tokens where we can get 1.68, 1.02, and 0.52 BLEU increases compared with baseline, respectively.Further analyses show that our method can also improve the lexical diversity of translation.

Graph embedding (GE) methods embed nodes (and/or edges) in graph into a low-dimensional semantic space, and have shown its effectiveness in modeling multi-relational data.However, existing GE models are not practical in real-world applications since it overlooked the streaming nature of incoming data.To address this issue, we study the problem of continual graph representation learning which aims to continually train a graph embedding model on new data to learn incessantly emerging multi-relational data while avoiding catastrophically forgetting old learned knowledge.Moreover, we propose a disentangle-based continual graph representation learning (DiC-GRL) framework inspired by the human's ability to learn procedural knowledge.The experimental results show that DiCGRL could effectively alleviate the catastrophic forgetting problem and outperform state-of-the-art continual learning models.* This work is done when Xiaoyu Kou was interning at Pattern Recognition Center, WeChat AI, Tencent Inc, China !"#"$% &'"(" )*$+,--, &'"(" )"-*" .//&'"(" .//,01/+"( 2#,3*4,/5 6+, 7/*5,4 85"5,3 85"5, 9: ;"<"** ="<>,# ?#"3,# @.B9'*/39/ ?*#35 ="4> 9: 78 @+*$"C9

Neural models have achieved remarkable success on relation extraction (RE) benchmarks.However, there is no clear understanding which type of information affects existing RE models to make decisions and how to further improve the performance of these models.To this end, we empirically study the effect of two main information sources in text: textual context and entity mentions (names).We find that (i) while context is the main source to support the predictions, RE models also heavily rely on the information from entity mentions, most of which is type information, and (ii) existing datasets may leak shallow heuristics via entity mentions and thus contribute to the high performance on RE benchmarks.Based on the analyses, we propose an entity-masked contrastive pre-training framework for RE to gain a deeper understanding on both textual context and type information while avoiding rote memorization of entities or use of superficial cues in mentions.We carry out extensive experiments to support our views, and show that our framework can improve the effectiveness and robustness of neural models in different RE scenarios.All the code and datasets are released at https://github.com/thunlp/

Being able to perform in-depth chat with humans in a closed domain is a precondition before an open-domain chatbot can ever be claimed.In this work, we take a close look at the movie domain and present a large-scale high-quality corpus with fine-grained annotations in hope of pushing the limit of moviedomain chatbots.We propose a unified, readily scalable neural approach which reconciles all subtasks like intent prediction and knowledge retrieval.The model is first pretrained on the huge general-domain data, then finetuned on our corpus.We show this simple neural approach trained on high-quality data is able to outperform commercial systems replying on complex rules.On both the static and interactive tests, we find responses generated by our system exhibits remarkably good engagement and sensibleness close to human-written ones.We further analyze the limits of our work and point out potential directions for future work 1 .

Xiaozhi Wang, Ziqi Wang, Xu Han, Wangyi Jiang, Rong Han, Zhiyuan Liu, Juanzi Li, Peng Li, Yankai Lin, Jie Zhou. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2020.

Transformer models (Vaswani et al., 2017) achieve remarkable success in Neural Machine Translation.Many efforts have been devoted to deepening the Transformer by stacking several units (i.e., a combination of Multihead Attentions and FFN) in a cascade, while the investigation over multiple parallel units draws little attention.In this paper, we propose the Multi-Unit TransformErs (MUTE), which aim to promote the expressiveness of the Transformer by introducing diverse and complementary units.Specifically, we use several parallel units and show that modeling with multiple units improves model performance and introduces diversity.Further, to better leverage the advantage of the multi-unit setting, we design biased module and sequential dependency that guide and encourage complementariness among different units.Experimental results on three machine translation tasks, the NIST Chinese-to-English, WMT'14 English-to-German and WMT'18 Chinese-to-English, show that the MUTE models significantly outperform the Transformer-Base, by up to +1.52, +1.90 and +1.10 BLEU points, with only a mild drop in inference speed (about 3.1%).In addition, our methods also surpass the Transformer-Big model, with only 54% of its parameters.These results demonstrate the effectiveness of the MUTE, as well as its efficiency in both the inference process and parameter usage. 1

More than 300 million people have been affected by depression all over the world. Due to the medical equipment and knowledge limitations, most of them are not diagnosed at the early stages. Recent work attempts to use social media to detect depression since the patterns of opinions and thoughts expression of the posted text and images, can reflect users' mental state to some extent. In this work, we design a system dubbed SenseMood to demonstrate that the users with depression can be efficiently detected and analyzed by using proposed system. A deep visual-textual multimodal learning approach has been proposed to reveal the psychological state of the users on social networks. The posted images and tweets data from users with/without depression on Twitter have been collected and used for depression detection. CNN-based classifier and Bert are applied to extract the deep features from the pictures and text posted by users respectively. Then visual and textual features are combined to reflect the emotional expression of users. Finally our system classifies the users with depression and normal users through a neural network and the analysis report is generated automatically.

Multimodal machine translation (MMT), which mainly focuses on enhancing text-only translation with visual features, has attracted considerable attention from both computer vision and natural language processing communities. Most current MMT models resort to attention mechanism, global context modeling or multimodal joint representation learning to utilize visual features. However, the attention mechanism lacks sufficient semantic interactions between modalities while the other two provide fixed visual context, which is unsuitable for modeling the observed variability when generating translation. To address the above issues, in this paper, we propose a novel Dynamic Context-guided Capsule Network (DCCN) for MMT. Specifically, at each timestep of decoding, we first employ the conventional source-target attention to produce a timestep-specific source-side context vector. Next, DCCN takes this vector as input and uses it to guide the iterative extraction of related visual features via a context-guided dynamic routing mechanism. Particularly, we represent the input image with global and regional visual features, we introduce two parallel DCCNs to model multimodal context vectors with visual features at different granularities. Finally, we obtain two multimodal context vectors, which are fused and incorporated into the decoder for the prediction of the target word. Experimental results on the Multi30K dataset of English-to-German and English-to-French translation demonstrate the superiority of DCCN. Our code is available on https://github.com/DeepLearnXMU/MM-DCCN.

Document Grounded Conversations is a task to generate dialogue responses when chatting about the content of a given document. Obviously, document knowledge plays a critical role in Document Grounded Conversations, while existing dialogue models do not exploit this kind of knowledge effectively enough. In this paper, we propose a novel Transformer-based architecture for multi-turn document grounded conversations. In particular, we devise an Incremental Transformer to encode multi-turn utterances along with knowledge in related documents. Motivated by the human cognitive process, we design a two-pass decoder (Deliberation Decoder) to improve context coherence and knowledge correctness. Our empirical study on a real-world Document Grounded Dataset proves that responses generated by our model significantly outperform competitive baselines on both context coherence and knowledge relevance.

Rhetoric is a vital element in modern poetry, and plays an essential role in improving its aesthetics.However, to date, it has not been considered in research on automatic poetry generation.In this paper, we propose a rhetorically controlled encoder-decoder for modern Chinese poetry generation.Our model relies on a continuous latent variable as a rhetoric controller to capture various rhetorical patterns in an encoder, and then incorporates rhetoricbased mixtures while generating modern Chinese poetry.For metaphor and personification, an automated evaluation shows that our model outperforms state-of-the-art baselines by a substantial margin, while a human evaluation shows that our model generates better poems than baseline methods in terms of fluency, coherence, meaningfulness, and rhetorical aesthetics.

Current state-of-the-art systems for the sequence labeling tasks are typically based on the family of Recurrent Neural Networks (RNNs).However, the shallow connections between consecutive hidden states of RNNs and insufficient modeling of global information restrict the potential performance of those models.In this paper, we try to address these issues, and thus propose a Global Context enhanced Deep Transition architecture for sequence labeling named GCDT.We deepen the state transition path at each position in a sentence, and further assign every token with a global representation learned from the entire sentence.Experiments on two standard sequence labeling tasks show that, given only training data and the ubiquitous word embeddings (Glove), our GCDT achieves 91.96 F 1 on the CoNLL03 NER task and 95.43 F 1 on the CoNLL2000 Chunking task, which outperforms the best reported results under the same settings.Furthermore, by leveraging BERT as an additional resource, we establish new stateof-the-art results with 93.47 F 1 on NER and 97.30F 1 on Chunking 1 .

In this paper, we focus on the task of finegrained text sentiment transfer (FGST).This task aims to revise an input sequence to satisfy a given sentiment intensity, while preserving the original semantic content.Different from conventional sentiment transfer task that only reverses the sentiment polarity (positive/negative) of text, the FTST task requires more nuanced and fine-grained control of sentiment.To remedy this, we propose a novel Seq2SentiSeq model.Specifically, the numeric sentiment intensity value is incorporated into the decoder via a Gaussian kernel layer to finely control the sentiment intensity of the output.Moreover, to tackle the problem of lacking parallel data, we propose a cycle reinforcement learning algorithm to guide the model training.In this framework, the elaborately designed rewards can balance both sentiment transformation and content preservation, while not requiring any ground truth output.Experimental results show that our approach can outperform existing methods by a large margin in both automatic evaluation and human evaluation.Our code and data, including outputs of all baselines and our model are available at https://github.com/luofuli/ Fine-grained-Sentiment-Transfer. 1

Table -to-text generation aims to translate the structured data into the unstructured text.Most existing methods adopt the encoder-decoder framework to learn the transformation, which requires large-scale training samples.However, the lack of large parallel data is a major practical problem for many domains.In this work, we consider the scenario of low resource table-to-text generation, where only limited parallel data is available.We propose a novel model to separate the generation into two stages: key fact prediction and surface realization.It first predicts the key facts from the tables, and then generates the text with the key facts.The training of key fact prediction needs much fewer annotated data, while surface realization can be trained with pseudo parallel corpus.We evaluate our model on a biography generation dataset.Our model can achieve 27.34 BLEU score with only 1, 000 parallel data, while the baseline model only obtain the performance of 9.71 BLEU score. 1

Non-Autoregressive Transformer (NAT) aims to accelerate the Transformer model through discarding the autoregressive mechanism and generating target words independently, which fails to exploit the target sequential information.Over-translation and under-translation errors often occur for the above reason, especially in the long sentence translation scenario.In this paper, we propose two approaches to retrieve the target sequential information for NAT to enhance its translation ability while preserving the fast-decoding property.Firstly, we propose a sequence-level training method based on a novel reinforcement algorithm for NAT (Reinforce-NAT) to reduce the variance and stabilize the training procedure.Secondly, we propose an innovative Transformer decoder named FS-decoder to fuse the target sequential information into the top layer of the decoder.Experimental results on three translation tasks show that the Reinforce-NAT surpasses the baseline NAT system by a significant margin on BLEU without decelerating the decoding speed and the FS-decoder achieves comparable translation performance to the autoregressive Transformer with considerable speedup.

Recent research has achieved impressive results in single-turn dialogue modelling. In the multi-turn setting, however, current models are still far from satisfactory. One major challenge is the frequently occurred coreference and information omission in our daily conversation, making it hard for machines to understand the real intention. In this paper, we propose rewriting the human utterance as a pre-process to help multi-turn dialgoue modelling. Each utterance is first rewritten to recover all coreferred and omitted information. The next processing steps are then performed based on the rewritten utterance. To properly train the utterance rewriter, we collect a new dataset with human annotations and introduce a Transformer-based utterance rewriting architecture using the pointer network. We show the proposed architecture achieves remarkably good performance on the utterance rewriting task. The trained utterance rewriter can be easily integrated into online chatbots and brings general improvement over different domains.

Yuan Yao, Deming Ye, Peng Li, Xu Han, Yankai Lin, Zhenghao Liu, Zhiyuan Liu, Lixin Huang, Jie Zhou, Maosong Sun. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. 2019.

Tianyu Gao, Xu Han, Hao Zhu, Zhiyuan Liu, Peng Li, Maosong Sun, Jie Zhou. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Yunlong Liang, Fandong Meng, Jinchao Zhang, Jinan Xu, Yufeng Chen, Jie Zhou. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Yijin Liu, Fandong Meng, Jinchao Zhang, Jie Zhou, Yufeng Chen, Jinan Xu. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Qiu Ran, Yankai Lin, Peng Li, Jie Zhou, Zhiyuan Liu. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Xiaozhi Wang, Ziqi Wang, Xu Han, Zhiyuan Liu, Juanzi Li, Peng Li, Maosong Sun, Jie Zhou, Xiang Ren. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Zhengxin Yang, Jinchao Zhang, Fandong Meng, Shuhao Gu, Yang Feng, Jie Zhou. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Unsupervised text style transfer aims to transfer the underlying style of text but keep its main content unchanged without parallel data. Most existing methods typically follow two steps: first separating the content from the original style, and then fusing the content with the desired style. However, the separation in the first step is challenging because the content and style interact in subtle ways in natural language. Therefore, in this paper, we propose a dual reinforcement learning framework to directly transfer the style of the text via a one-step mapping model, without any separation of content and style. Specifically, we consider the learning of the source-to-target and target-to-source mappings as a dual task, and two rewards are designed based on such a dual structure to reflect the style accuracy and content preservation, respectively. In this way, the two one-step mapping models can be trained via reinforcement learning, without any use of parallel data. Automatic evaluations show that our model outperforms the state-of-the-art systems by a large margin, especially with more than 10 BLEU points improvement averaged on two benchmark datasets. Human evaluations also validate the effectiveness of our model in terms of style accuracy, content preservation and fluency. Our code and data, including outputs of all baselines and our model are available at https://github.com/luofuli/DualRL.

Deep Neural Networks (DNNs) have provably enhanced the state-of-the-art Neural Machine Translation (NMT) with their capability in modeling complex functions and capturing complex linguistic structures.However NMT systems with deep architecture in their encoder or decoder RNNs often suffer from severe gradient diffusion due to the non-linear recurrent activations, which often make the optimization much more difficult.To address this problem we propose novel linear associative units (LAU) to reduce the gradient propagation length inside the recurrent unit.Different from conventional approaches (LSTM unit and GRU), LAUs utilizes linear associative connections between input and output of the recurrent unit, which allows unimpeded information flow through both space and time direction.The model is quite simple, but it is surprisingly effective.Our empirical study on Chinese-English translation shows that our model with proper configuration can improve by 11.7 BLEU upon Groundhog and the best reported results in the same setting.On WMT14 English-German task and a larger WMT14 English-French task, our model achieves comparable results with the state-of-the-art.

This paper proposes three distortion models to explicitly incorporate the word reordering knowledge into attention-based Neural Machine Translation (NMT) for further improving translation performance.Our proposed models enable attention mechanism to attend to source words regarding both the semantic requirement and the word reordering penalty.Experiments on Chinese-English translation show that the approaches can improve word alignment quality and achieve significant translation improvements over a basic attention-based N-MT by large margins.Compared with previous works on identical corpora, our system achieves the state-of-the-art performance on translation quality.

The encoder-decoder neural framework is widely employed for Neural Machine Translation (NMT) with a single encoder to represent the source sentence and a single decoder to generate target words. The translation performance heavily relies on the representation ability of the encoder and the generation ability of the decoder. To further enhance NMT, we propose to extend the original encoder-decoder framework to a novel one, which has multiple encoders and decoders (ME-MD). Through this way, multiple encoders extract more diverse features to represent the source sequence and multiple decoders capture more complicated translation knowledge. Our proposed ME-MD framework is convenient to integrate heterogeneous encoders and decoders with multiple depths and multiple types. Experiment on Chinese-English translation task shows that our ME-MD system surpasses the state-of-the-art NMT system by 2.1 BLEU points and surpasses the phrase-based Moses by 7.38 BLEU points. Our framework is general and can be applied to other sequence to sequence tasks.