Yuming Yang, Mingyoung Lai, Wanxu Zhao, Xiaoran Fan, Zhiheng Xi, Mingqi Wu, Chiyue Huang, Jun Zhao, Haijun Lv, Jian Tong, Yunhua Zhou, Yicheng Zou, Qipeng Guo, Tao Gui, Qi Zhang, Xuanjing Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Zhi Chen, Qiguang Chen, Libo Qin, Qipeng Guo, Haijun Lv, Yicheng Zou, Hang Yan, Kai Chen, Dahua Lin. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Scaling law builds the relationship between training computation and validation loss, enabling researchers to effectively predict the loss trending of models across different levels of computation. However, a gap still remains between validation loss and the model’s downstream capabilities, making it untrivial to apply scaling law to direct performance prediction for downstream tasks. The loss typically represents a cumulative penalty for predicted tokens, which are implicitly considered to have equal importance. Nevertheless, our studies have shown evidence that when considering different training data distributions, we cannot directly model the relationship between downstream capability and computation or token loss. To bridge the gap between validation loss and downstream task capabilities, in this work, we introduce Capability Salience Vector, which decomposes the overall loss and assigns different importance weights to tokens to assess a specific meta-capability, aligning the validation loss with downstream task performance in terms of the model’s capabilities. Experiments on various popular benchmarks demonstrate that our proposed Capability Salience Vector could significantly improve the predictability of language model performance on downstream tasks.

Zhiheng Xi, Yiwen Ding, Wenxiang Chen, Boyang Hong, Honglin Guo, Junzhe Wang, Xin Guo, Dingwen Yang, Chenyang Liao, Wei He, Songyang Gao, Lu Chen, Rui Zheng, Yicheng Zou, Tao Gui, Qi Zhang, Xipeng Qiu, Xuanjing Huang, Zuxuan Wu, Yu-Gang Jiang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

We offer a novel perspective on reward modeling by formulating it as a policy discriminator, which quantifies the difference between two policies to generate a reward signal, guiding the training policy towards a target policy with desired behaviors. Based on this conceptual insight, we propose a scalable pre-training method named POLicy DiscriminAtive LeaRning (POLAR), which trains a reward model (RM) to discern identical policies and discriminate different ones. Unlike traditional reward modeling methods relying on absolute preferences, POLAR captures the relative difference between one policy and an arbitrary target policy, which is a scalable, high-level optimization objective suitable for modeling generic ranking relationships. Leveraging the POLAR pre-training paradigm, we present a series of RMs with parameter scales from 1.8B to 7B. Empirical results show that POLAR substantially outperforms traditional non-pre-trained methods, significantly enhancing RM performance. For instance, POLAR-7B could improve preference accuracy from 54.8% to 81.0% on STEM tasks and from 57.9% to 85.5% on creative writing tasks compared to SOTA baselines. POLAR also shows robust generalization capabilities in RLHF using Reinforcement Fine-tuning (RFT), providing reliable reward signals and markedly enhancing policy performance—improving LLaMa3.1-8B from an average of 47.36% to 56.33% and Qwen2.5-32B from 64.49% to 70.47% on 20 benchmarks. Moreover, scaling experiments reveal a clear power-law relationship between computation and performance, supported by linear correlation coefficients approaching 0.99. The impressive performance, strong generalization, and scaling properties suggest that POLAR is a promising direction for developing general and strong reward models.

The success of AI assistants based on Language Models (LLMs) hinges on Reinforcement Learning from Human Feedback (RLHF) to comprehend and align with user intentions. However, traditional alignment algorithms, such as PPO, are hampered by complex annotation and training requirements. This reliance limits the applicability of RLHF and hinders the development of professional assistants tailored to diverse human preferences. In this work, we introduce Linear Alignment, a novel algorithm that aligns language models with human preferences in one single inference step, eliminating the reliance on data annotation and model training. Linear alignment incorporates a new parameterization for policy optimization under divergence constraints, which enables the extraction of optimal policy in a closed-form manner and facilitates the direct estimation of the aligned response. Extensive experiments on both general and personalized preference datasets demonstrate that linear alignment significantly enhances the performance and efficiency of LLM alignment across diverse scenarios.

Dialogue summarization aims to condense the lengthy dialogue into a concise summary, and has recently achieved significant progress.However, the result of existing methods is still far from satisfactory.Previous works indicated that omission is a major factor in affecting the quality of summarization, but few of them have further explored the omission problem, such as how omission affects summarization results and how to detect omission, which is critical for reducing omission and improving summarization quality.Moreover, analyzing and detecting omission relies on summarization datasets with omission labels (i.e., which dialogue utterances are omitted in the summarization), which are not available in the current literature.In this paper, we propose the OLDS dataset, which provides high-quality Omission Labels for Dialogue Summarization.By analyzing this dataset, we find that a large improvement in summarization quality can be achieved by providing ground-truth omission labels for the summarization model to recover omission information, which demonstrates the importance of omission detection for omission mitigation in dialogue summarization.Therefore, we formulate an omission detection task and demonstrate our proposed dataset can support the training and evaluation of this task well.We also call for research action on omission detection based on our proposed datasets.Our dataset and codes are publicly available 1 .

Xiao Wang, Shihan Dou, Limao Xiong, Yicheng Zou, Qi Zhang, Tao Gui, Liang Qiao, Zhanzhan Cheng, Xuanjing Huang. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.

Large pre-trained language models (PLMs) have demonstrated superior performance in industrial applications. Recent studies have explored parameter-efficient PLM tuning, which only updates a small amount of task-specific parameters while achieving both high efficiency and comparable performance against standard fine-tuning. However, all these methods ignore the inefficiency problem caused by the task-specific output layers, which is inflexible for us to re-use PLMs and introduces non-negligible parameters. In this work, we focus on the text classification task and propose plugin-tuning, a framework that further improves the efficiency of existing parameter-efficient methods with a unified classifier. Specifically, we re-formulate both token and sentence classification tasks into a unified language modeling task, and map label spaces of different tasks into the same vocabulary space. In this way, we can directly re-use the language modeling heads of PLMs, avoiding introducing extra parameters for different tasks. We conduct experiments on six classification benchmarks. The experimental results show that plugin-tuning can achieve comparable performance against fine-tuned PLMs, while further saving around 50% parameters on top of other parameter-efficient methods.

Multimodal machine translation (MMT) aims to improve translation quality by equipping the source sentence with its corresponding image.Despite the promising performance, MMT models still suffer the problem of input degradation: models focus more on textual information while visual information is generally overlooked.In this paper, we endeavor to improve MMT performance by increasing visual awareness from an information theoretic perspective.In detail, we decompose the informative visual signals into two parts: source-specific information and target-specific information.We use mutual information to quantify them and propose two methods for objective optimization to better leverage visual signals.Experiments on two datasets demonstrate that our approach can effectively enhance the visual awareness of MMT model and achieve superior results against strong baselines.

Sentence scoring aims at measuring the likelihood score of a sentence and is widely used in many natural language processing scenarios, like reranking, which is to select the best sentence from multiple candidates. Previous works on sentence scoring mainly adopted either causal language modeling (CLM) like GPT or masked language modeling (MLM) like BERT, which have some limitations: 1) CLM only utilizes unidirectional information for the probability estimation of a sentence without considering bidirectional context, which affects the scoring quality; 2) MLM can only estimate the probability of partial tokens at a time and thus requires multiple forward passes to estimate the probability of the whole sentence, which incurs large computation and time cost. In this paper, we propose \textit{Transcormer} -- a Transformer model with a novel \textit{sliding language modeling} (SLM) for sentence scoring. Specifically, our SLM adopts a triple-stream self-attention mechanism to estimate the probability of all tokens in a sentence with bidirectional context and only requires a single forward pass. SLM can avoid the limitations of CLM (only unidirectional context) and MLM (multiple forward passes) and inherit their advantages, and thus achieve high effectiveness and efficiency in scoring. Experimental results on multiple tasks demonstrate that our method achieves better performance than other language modelings.

Automatic chat summarization can help people quickly grasp important information from numerous chat messages. Unlike conventional documents, chat logs usually have fragmented and evolving topics. In addition, these logs contain a quantity of elliptical and interrogative sentences, which make the chat summarization highly context dependent. In this work, we propose a novel unsupervised framework called RankAE to perform chat summarization without employing manually labeled data. RankAE consists of a topic-oriented ranking strategy that selects topic utterances according to centrality and diversity simultaneously, as well as a denoising auto-encoder that is carefully designed to generate succinct but context-informative summaries based on the selected utterances. To evaluate the proposed method, we collect a large-scale dataset of chat logs from a customer service environment and build an annotated set only for model evaluation. Experimental results show that RankAE significantly outperforms other unsupervised methods and is able to generate high-quality summaries in terms of relevance and topic coverage.

In a customer service system, dialogue summarization can boost service efficiency by automatically creating summaries for long spoken dialogues in which customers and agents try to address issues about specific topics. In this work, we focus on topic-oriented dialogue summarization, which generates highly abstractive summaries that preserve the main ideas from dialogues. In spoken dialogues, abundant dialogue noise and common semantics could obscure the underlying informative content, making the general topic modeling approaches difficult to apply. In addition, for customer service, role-specific information matters and is an indispensable part of a summary. To effectively perform topic modeling on dialogues and capture multi-role information, in this work we propose a novel topic-augmented two-stage dialogue summarizer (TDS) jointly with a saliency-aware neural topic model (SATM) for topic-oriented summarization of customer service dialogues. Comprehensive studies on a real-world Chinese customer service dataset demonstrated the superiority of our method against several strong baselines.

Aspect-based sentiment analysis aims to identify the sentiment polarity of a specific aspect in product reviews.We notice that about 30% of reviews do not contain obvious opinion words, but still convey clear human-aware sentiment orientation, which is known as implicit sentiment.However, recent neural networkbased approaches paid little attention to implicit sentiment entailed in the reviews.To overcome this issue, we adopt Supervised Contrastive Pre-training on large-scale sentimentannotated corpora retrieved from in-domain language resources.By aligning the representation of implicit sentiment expressions to those with the same sentiment label, the pretraining process leads to better capture of both implicit and explicit sentiment orientation towards aspects in reviews.Experimental results show that our method achieves state-of-theart performance on SemEval2014 benchmarks, and comprehensive analysis validates its effectiveness on learning implicit sentiment.

Human dialogue contains evolving concepts, and speakers naturally associate multiple concepts to compose a response.However, current dialogue models with the seq2seq framework lack the ability to effectively manage concept transitions and can hardly introduce multiple concepts to responses in a sequential decoding manner.To facilitate a controllable and coherent dialogue, in this work, we devise a conceptguided non-autoregressive model (CG-nAR) for open-domain dialogue generation.The proposed model comprises a multi-concept planning module that learns to identify multiple associated concepts from a concept graph and a customized Insertion Transformer that performs concept-guided non-autoregressive generation to complete a response.The experimental results on two public datasets show that CG-nAR can produce diverse and coherent responses, outperforming state-of-the-art baselines in both automatic and human evaluations with substantially faster inference speed.

With the rapid increase in the volume of dialogue data from daily life, there is a growing demand for dialogue summarization.Unfortunately, training a large summarization model is generally infeasible due to the inadequacy of dialogue data with annotated summaries.Most existing works for low-resource dialogue summarization directly pretrain models in other domains, e.g., the news domain, but they generally neglect the huge difference between dialogues and conventional articles.To bridge the gap between out-of-domain pretraining and indomain fine-tuning, in this work, we propose a multi-source pretraining paradigm to better leverage the external summary data.Specifically, we exploit large-scale in-domain nonsummary data to separately pretrain the dialogue encoder and the summary decoder.The combined encoder-decoder model is then pretrained on the out-of-domain summary data using adversarial critics, aiming to facilitate domain-agnostic summarization.The experimental results on two public datasets show that with only limited training data, our approach achieves competitive performance and generalizes well in different dialogue scenarios.

Tao Gui, Yicheng Zou, Qi Zhang, Minlong Peng, Jinlan Fu, Zhongyu Wei, Xuanjing Huang. 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.

Attention mechanisms have been leveraged for sentiment classification tasks because not all words have the same importance. However, most existing attention models did not take full advantage of sentiment lexicons, which provide rich sentiment information and play a critical role in sentiment analysis. To achieve the above target, in this work, we propose a novel lexicon-based supervised attention model (LBSA), which allows a recurrent neural network to focus on the sentiment content, thus generating sentiment-informative representations. Compared with general attention models, our model has better interpretability and less noise. Experimental results on three large-scale sentiment classification datasets showed that the proposed method outperforms previous methods.