Wen Luo, Guangyue Peng, Wei Li, Shaohang Wei, Feifan Song, Liang Wang, Nan Yang, Xingxing Zhang, Jing Jin, Furu Wei, Houfeng Wang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Large Language Models (LLMs) are increasingly required to generate text that is both factually accurate and diverse across various openended applications.However, current stochastic decoding methods struggle to balance such objectives.We introduce Dynamic Focus Decoding (DFD), a novel plug-and-play stochastic approach that resolves this trade-off without requiring additional data, knowledge, or models.DFD adaptively adjusts the decoding focus based on distributional differences across layers, leveraging the modular and hierarchical nature of factual knowledge within LLMs.This dynamic adjustment improves factuality in knowledge-intensive decoding steps and promotes diversity in less knowledge-reliant steps.DFD can be easily integrated with existing decoding methods, enhancing both factuality and diversity with minimal computational overhead.Extensive experiments across seven datasets demonstrate that DFD significantly improves performance, providing a scalable and efficient solution for open-ended text generation. 1

Supervised fine-tuning (SFT) enables large language models to align with training data for better performance in many aspects.Nevertheless, the gap between the distribution of current datasets from human annotations or model generations and the real-world data distribution heavily limits the capacities and potentials of models.As a result, we propose a new SFT technique, ATLANTIS, to bridge the gap.We adopt importance sampling to estimate the optimal data distribution in the real world from existing training datasets because the former is hard to sample from.Furthermore, we introduce an extra small model and reference model to estimate the sampling ratio through the probability gap between them.We evaluate our method with benchmarks in knowledge & understanding and preference aspects.The experiment results prove that ATLANTIS can bring consistent and significant improvements to models' performance.What's more, our method can be flexibly transferred among models with different structures.Our analyses demonstrate that our method is well-compatible with other SFT techniques to further enhance models' capacities and has great potential to be combined with existing training frameworks.

Recent advancements in large language models (LLMs) have led to significant breakthroughs in mathematical reasoning capabilities. However, existing benchmarks like GSM8K or MATH are now being solved with high accuracy (e.g., OpenAI o1 achieves 94.8% on MATH dataset), indicating their inadequacy for truly challenging these models. To bridge this gap, we propose a comprehensive and challenging benchmark specifically designed to assess LLMs' mathematical reasoning at the Olympiad level. Unlike existing Olympiad-related benchmarks, our dataset focuses exclusively on mathematics and comprises a vast collection of 4428 competition-level problems with rigorous human annotation. These problems are meticulously categorized into over 33 sub-domains and span more than 10 distinct difficulty levels, enabling a holistic assessment of model performance in Olympiad-mathematical reasoning. Furthermore, we conducted an in-depth analysis based on this benchmark. Our experimental results show that even the most advanced models, OpenAI o1-mini and OpenAI o1-preview, struggle with highly challenging Olympiad-level problems, with 60.54% and 52.55% accuracy, highlighting significant challenges in Olympiad-level mathematical reasoning.

Feifan Song, Yuxuan Fan, Xin Zhang, Peiyi Wang, Houfeng Wang. 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.

Temporal reasoning is pivotal for Large Language Models (LLMs) to comprehend the real world. However, existing works neglect the real-world challenges for temporal reasoning: (1) intensive temporal information, (2) fast-changing event dynamics, and (3) complex temporal dependencies in social interactions. To bridge this gap, we propose a multi-level benchmark TimE, designed for temporal reasoning in real-world scenarios. TimE consists of 38,522 QA pairs, covering 3 levels with 11 fine-grained sub-tasks. This benchmark encompasses 3 sub-datasets reflecting different real-world challenges: TimE-Wiki, TimE-News, and TimE-Dial. We conduct extensive experiments on reasoning models and non-reasoning models. And we conducted an in-depth analysis of temporal reasoning performance across diverse real-world scenarios and tasks, and summarized the impact of test-time scaling on temporal reasoning capabilities. Additionally, we release TimE-Lite, a human-annotated subset to foster future research and standardized evaluation in temporal reasoning.

Large language models (LLMs) often contain misleading content, emphasizing the need to align them with human values to ensure secure AI systems. Reinforcement learning from human feedback (RLHF) has been employed to achieve this alignment. However, it encompasses two main drawbacks: (1) RLHF exhibits complexity, instability, and sensitivity to hyperparameters in contrast to SFT. (2) Despite massive trial-and-error, multiple sampling is reduced to pair-wise contrast, thus lacking contrasts from a macro perspective. In this paper, we propose Preference Ranking Optimization (PRO) as an efficient SFT algorithm to directly fine-tune LLMs for human alignment. PRO extends the pair-wise contrast to accommodate preference rankings of any length. By iteratively contrasting candidates, PRO instructs the LLM to prioritize the best response while progressively ranking the rest responses. In this manner, PRO effectively transforms human alignment into aligning the probability ranking of n responses generated by LLM with the preference ranking of humans towards these responses. Experiments have shown that PRO outperforms baseline algorithms, achieving comparable results to ChatGPT and human responses through automatic-based, reward-based, GPT-4, and human evaluations.

Alignment with human preference prevents large language models (LLMs) from generating misleading or toxic content while requiring high-cost human feedback. Assuming resources of human annotation are limited, there are two different ways of allocating considered: more diverse PROMPTS or more diverse RESPONSES to be labeled. Nonetheless, a straightforward comparison between their impact is absent. In this work, we first control the diversity of both sides according to the number of samples for fine-tuning, which can directly reflect their influence. We find that instead of numerous prompts, more responses but fewer prompts better trigger LLMs for human alignment. Additionally, the concept of diversity for prompts can be more complex than responses that are typically quantified by single digits. Consequently, a new formulation of prompt diversity is proposed, further implying a linear correlation with the final performance of LLMs after fine-tuning. We also leverage it on data augmentation and conduct experiments to show its effect on different algorithms.

ChatGPT has demonstrated impressive capabilities in building conversations. However, for Spoken Language Understanding (SLU) with multiple intents, traditional approaches where Intent Detection and Slot Filling are jointly modeled with distinct formulations hinder networks from effectively extracting shared features. In this work, we describe a Prompt-based SLU (PromptSLU) framework, to intuitively unify two sub-tasks into the same form for a common pre-trained model. Specifically, variable intents are predicted first, then naturally embedded into prompts to guide slot-value inference from a semantic perspective. Furthermore, we are inspired by multi-task learning to introduce an auxiliary sub-task and a concise general objective, which helps to learn relationships among provided labels. Experiment results show that our framework outperforms several competitive baselines on two datasets. The source code is available at https://github.com/F2-Song/PromptSLU.

Recent research has demonstrated that Large Language Models (LLMs) can enhance their capabilities by utilizing external tools.However, three pivotal questions remain unanswered: (1) How effective are current LLMs in utilizing tools?(2) How can we enhance LLMs' ability to utilize tools?(3) What obstacles need to be overcome to leverage tools?To address these questions, we introduce API-Bank, a groundbreaking benchmark, specifically designed for tool-augmented LLMs.For the first question, we develop a runnable evaluation system consisting of 73 API tools.We annotate 314 tool-use dialogues with 753 API calls to assess the existing LLMs' capabilities in planning, retrieving, and calling APIs.For the second question, we construct a comprehensive training set containing 1,888 tool-use dialogues from 2,138 APIs spanning 1,000 distinct domains.Using this dataset, we train Lynx, a tool-augmented LLM initialized from Alpaca.Experimental results demonstrate that GPT-3.5 exhibits improved tool utilization compared to GPT-3, while GPT-4 excels in planning.However, there is still significant potential for further improvement.Moreover, Lynx surpasses Alpaca's tool utilization performance by more than 26 pts and approaches the effectiveness of GPT-3.5.Through error analysis, we highlight the key challenges for future research in this field to answer the third question 1 .

Proactive dialogue system is able to lead the conversation to a goal topic and has advantaged potential in bargain, persuasion, and negotiation. Current corpus-based learning manner limits its practical application in real-world scenarios. To this end, we contribute to advancing the study of the proactive dialogue policy to a more natural and challenging setting, i.e., interacting dynamically with users. Further, we call attention to the non-cooperative user behavior - the user talks about off-path topics when he/she is not satisfied with the previous topics introduced by the agent. We argue that the targets of reaching the goal topic quickly and maintaining a high user satisfaction are not always converged, because the topics close to the goal and the topics user preferred may not be the same. Towards this issue, we propose a new solution named I-Pro that can learn Proactive policy in the Interactive setting. Specifically, we learn the trade-off via a learned goal weight, which consists of four factors (dialogue turn, goal completion difficulty, user satisfaction estimation, and cooperative degree). The experimental results demonstrate I-Pro significantly outperforms baselines in terms of effectiveness and interpretability.