In this paper, we introduce Predicate-State Video Reasoning Segmentation (PS-VRS), where a query defines a dynamic predicate P(o,t) over objects and time rather than a fixed object identity. Existing methods typically treat the target as permanently active once identified, leading to over-emission of masks during vacant intervals and failure to handle target switching. To address this, we construct PS-Bench, a large-scale benchmark with 2,077 videos and 16,250 queries annotated with explicit active/vacant predicate states, temporal intervals, and pixel masks. We further propose Predicate-State Grounding (PSG), a State-Bind-Emit pipeline that: (1) predicts [ACTIVE]/[VACANT] state tokens; (2) aligns them to video frames via dense cross-modal similarity; and (3) emits masks via SAM2 only for active states. Experiments show that PSG outperforms existing baselines by +9.32 𝒥 ℱ on PS-Bench and significantly reduces false-positive mask emissions. Our results demonstrate the effectiveness of explicit predicate-state modeling for video reasoning segmentation with dynamic visual witnesses.

Vision-language-action (VLA) models for autonomous driving typically express intermediate reasoning entirely in text, which may discard decision-relevant visual information. We propose Latent Dreaming (LatDM), a latent–text reasoning framework that complements textual chain-of-thought with continuous latent states. LatDM uses latent blocks to preserve visually grounded spatial and temporal information, while text handles semantic abstraction and decision communication. A pretrained visual encoder provides latent supervision during training, and the model autonomously interleaves latent and textual reasoning at inference. Experiments on VLADBench, CODA-LM, and NuInstruct show consistent improvements over the underlying VLA backbone, while ablations confirm that the latent channel contributes beyond standard textual fine-tuning.

With the rapid development of large language models (LLMs), generative reward models (GRMs) have been widely adopted for reward modeling and evaluation. Previous studies have primarily focused on training specialized GRMs by optimizing them on preference datasets with the judgment correctness as supervision. While it’s widely accepted that GRMs with stronger problem-solving capabilities typically exhibit superior judgment abilities, we first identify a significant solve-to-judge gap when examining individual queries. Specifically, the solve-to-judge gap refers to the phenomenon where GRMs struggle to make correct judgments on some queries (14%-37%), despite being fully capable of solving them. In this paper, we propose the Solve-to-Judge (S2J) approach to address this problem. Specifically, S2J simultaneously leverages both the solving and judging capabilities on a single GRM’s output for supervision, explicitly linking the GRM’s problem-solving and evaluation abilities during model optimization, thereby narrowing the gap. Our comprehensive experiments demonstrate that S2J effectively reduces the solve-to-judge gap by 16.2%, thereby enhancing the model’s judgment performance by 5.8%. Notably, S2J achieves state-of-the-art (SOTA) performance among GRMs built on the same base model while utilizing a significantly smaller training dataset.

Reinforcement learning from verifiable rewards improves the reasoning ability of large language models, but often suffers from entropy collapse, in which increasingly concentrated policies reduce rollout diversity and useful learning signals. Existing remedies either constrain the RL objective (e.g., entropy regularization) or adjust sampling temperature during rollout collection, but these interventions remain external to the model parameters. We propose Temperature-Scaled On-Policy Self-Distillation (TS-OPSD), a lightweight policy reheating method that internalizes the exploratory effect of temperature into model parameters. Starting from an entropy-collapsed RL checkpoint, TS-OPSD constructs a self-teacher by applying high-temperature scaling to the model’s own logits, then distills the resulting smoother distribution back into the student. This policy reheating requires no external teacher, privileged data, or additional inference cost. Experiments on Qwen3-4B-Base and Qwen3-8B-Base show that policy reheating yields a stronger initialization for continued RL than both standard continued RL and rollout-level temperature reheating. Further analyses show that TS-OPSD mainly reduces output sharpness while preserving intermediate representations, top candidate sets, and reasoning capability. These results suggest that entropy restoration can serve as a simple post-collapse intervention for extending reasoning-oriented RL.

Despite the remarkable progress achieved by recent efficient methods in accelerating multimodal understanding, they still suffer from noticeable performance degradation. Their emphasis on the high compression ratio of a single visual clue and reliance on the heuristic pruning strategy with coarse attention alignment incurs a bottleneck on the information *capacity* and *density* of visual tokens. Addressing this limitation, we propose VEN-VL, a visual ensemble MoE framework for effective and efficient perception following the *enrich then compact* principle. Specifically, we first enrich the information capacity by unifying the visual representations of different perspectives, and then progressively compact it with adaptive routers in specialized visual experts to enhance the information density. Furthermore, we incorporate the reconstruction ability of vanilla structure via explicit visual supervision, facilitating crucial information preservation. Experimental results demonstrate our superiority in complex visual tasks with few information-condensed tokens, which effectively bridges the gap between performance and efficiency. Code will be available.

Evaluating multimodal large language models (MLLMs) on long-form scientific papers requires measuring not only answer correctness, but also whether answers are supported by traceable cross-modal evidence. Existing answer-only metrics and synthetic “Needle-In-A-Haystack” tests mainly assess answer matching or isolated retrieval, leaving evidence-grounded support under-examined. We introduce “Fish-in-the-Ocean” (FITO), an evidence-chain evaluation lens for native scientific documents, and instantiate it with SIN-Data, a scientific interleaved corpus that preserves text–figure coupling. Built on SIN-Data, SIN-Bench comprises four evidence-auditing tasks: evidence discovery, hypothesis verification, grounded QA, and evidence-anchored synthesis. We further propose “No Evidence, No Score”, which credits predictions only when linked to verifiable anchors and diagnoses evidence quality through matching, relevance, and logic. Experiments on 10 MLLMs show a clear correctness–grounding gap: Gemini-3-pro obtains the best average overall score (0.566), while GPT-5 reaches the highest SIN-QA answer accuracy (0.767) but lower evidence-aligned scores. Code and data are available at https://github.com/IIGROUP/sin-bench .

Supervised fine-tuning (SFT) is a fundamental post-training strategy to align Large Language Models (LLMs) with human intent. However, traditional SFT often ignores the one-to-many nature of language by forcing alignment with a single reference answer, leading to the model overfitting to non-core expressions. Although our empirical analysis suggests that introducing multiple reference answers can mitigate this issue, the prohibitive data and computational costs necessitate a strategic shift: prioritizing the mitigation of single-reference overfitting over the costly pursuit of answer diversity. To achieve this, we reveal the intrinsic connection between token probability and semantic importance: high-probability tokens carry the core logical framework, while low-probability tokens are mostly replaceable expressions. Based on this insight, we propose ProFit, which selectively masks low-probability tokens to prevent surface-level overfitting. Extensive experiments confirm that ProFit consistently outperforms traditional SFT baselines on general reasoning and mathematical benchmarks

The increasing adoption of large language models (LLMs) in software engineering necessitates rigorous security evaluation of their generated code. However, existing benchmarks often lack relevance to real-world AI-assisted programming scenarios, making them inadequate for assessing the practical security risks associated with AI-generated code in production environments. To address this gap, we introduce A.S.E (AI Code Generation Security Evaluation), a repository-level evaluation benchmark designed to closely mirror real-world AI programming tasks, offering a comprehensive and reliable framework for assessing the security of AI-generated code. Our evaluation of leading LLMs on A.S.E reveals several key findings. In particular, current LLMs still struggle with secure coding. The complexity in repository-level scenarios presents challenges for LLMs that typically perform well on snippet-level tasks. Moreover, a larger reasoning budget does not necessarily lead to better code generation. These observations offer valuable insights into the current state of AI code generation and help developers identify the most suitable models for practical tasks. They also lay the groundwork for refining LLMs to generate secure and efficient code in real-world applications.

The saturation of high-quality pre-training data has shifted research focus toward evolutionary systems capable of continuously generating novel artifacts, leading to the success of AlphaEvolve. However, the progress of such systems is hindered by the lack of rigorous, quantitative evaluation. To tackle this challenge, we introduce CreativeBench, a benchmark for evaluating machine creativity in code generation, grounded in a classical cognitive framework. Comprising two subsets – CreativeBench-Combo and CreativeBench-Explore – the benchmark targets combinatorial and exploratory creativity through an automated pipeline utilizing reverse engineering and self-play. By leveraging executable code, CreativeBench objectively distinguishes creativity from hallucination via a unified metric defined as the product of quality and novelty. Our analysis of state-of-the-art models reveals distinct behaviors: (1) scaling significantly improves combinatorial creativity but yields diminishing returns for exploration; (2) larger models exhibit “convergence-by-scaling,” becoming more correct but less divergent; and (3) reasoning capabilities primarily benefit constrained exploration rather than combination. Finally, we propose EvoRePE, a plug-and-play inference-time steering strategy that internalizes evolutionary search patterns to consistently enhance machine creativity.

Understanding how LLMs reason is hindered by a practical asymmetry: while their generated outputs are observable, the underlying reasoning patterns remain opaque. Relying on single probes, such as Mutual Information Peak (MIP) or Deep-Thinking Ratio (DTR), risks underestimating the genuine inferential structure. To response this deficiency, we present an Integrated, cross-Architecture Reasoning (IAR) framework, designed to provide a unified approach to LLM reasoning interpretability. Specifically, we first propose to use bandwidth-calibrated MIP coupled with Tukey IQR peak-detection to isolate reasoning-crucial tokens at the output layer. Second, we performed an overlap analysis between MIP-picked tokens and DTR-deep tokens to trace the cross-layer trajectories of those tokens. This also discloses whether reasoning-crucial tokens are computation-intensive as well, further facilitating to understand how reasoning patterns evolve across model layers. Finally, we apply a Jaccard stability metric over multi-domain problems to verify if the MIP-identified tokens are reasoning quality-guaranteed. Extensive experiments on six models across four domains (mathematics, code, logic, and common sense) demonstrate IAR’s generalizable interpretation capabilities across architectures. Our code is available at https://github.com/LeonardoMatthew/IAR.

Existing benchmarks for MLLM-generated web artifacts assess interaction through local evidence and miss the requirement-induced states and transitions that determine whether a page works. We introduce WebRISE, which compiles task requirements into Interaction Contract Graphs (ICGs) of observable states, user-intent transitions, and DOM/visual assertions for implementation-agnostic browser execution. WebRISE spans 442 tasks across five input modalities (Text, Markdown, Sketch, Image, Video), with 5,495 transitions and 5,271 requirement checks that separate user-stated functions from implicit product-level constraints. Across 14 MLLMs, even the strongest model reaches only 65.6% transition validity and 66.3% requirement coverage, and visual quality is no proxy for behavior (Qwen3.6-35B-A3B on Markdown: V=80.8 yet T=15.5). Video gives the strongest interaction signal (+10.6pp implicit coverage over Text), while implicit constraints persist; defect injection shows ICG-based scoring detects state errors at 2–16× the rate of checkpoint-style evaluation.

Universal multimodal retrieval (UMR) increasingly adopts multimodal large language models (MLLMs) as unified embedding backbones, but their strong retrieval performance comes at substantial inference cost. Existing methods typically rely on uniformly dense inference, where all input tokens are processed through the entire model and matched using the final-layer [EOS] representation. However, this paradigm overlooks two key forms of heterogeneity in multimodal retrieval: token contributions to the final retrieval embedding are highly uneven, and different queries require markedly different amounts of inference depth. To address this, we propose **Skim and Skip (SAS)**, a hierarchical adaptive inference framework for efficient multimodal retrieval. SAS first performs token-level evidence selection to preserve only the input information most relevant to the final retrieval embedding, and then performs depth-adaptive inference to determine whether the current representation is already sufficient for reliable matching. Experiments on 12 MMEB retrieval tasks show that SAS retains about 99% of the dense baseline’s average retrieval performance while reducing LLM-side FLOPs by up to **66.3%** and achieving up to **1.64×** end-to-end speedup. Our code is available at [https://github.com/GaoMengGladys/SAS](https://github.com/GaoMengGladys/SAS).

Between the first visible sign of danger and the moment an accident occurs, there is often a window where intervention remains possible. Video-capable multimodal large language models (MLLMs) could serve as always-on safety monitors that issue warnings during this window. Yet current benchmarks do not test this ability: they rely on static inputs, ignore timing precision, and omit false-positive measurement on safe scenes. We present PaSBench-Video, a 740-video benchmark with 481 risk videos and 259 no-risk videos across four domains: driving, healthcare, daily life, and industrial production. The benchmark is annotated with frame-level risk onset and accident boundaries. A model must observe the video causally and produce a warning that is both temporally calibrated and content-correct. Testing 13 MLLMs, we find that no model exceeds 20.0% on our strictest metric, and recall is tightly coupled with false-positive rate, with Pearson correlation 0.64: higher detection comes only at the cost of triggering warnings on the majority of safe clips. Performance splits sharply by domain: models achieve moderate recall at low false-positive rates in daily life, where risks are inherently anomalous, yet fire indiscriminately in driving, where routine and hazardous scenes look alike. These results indicate that current models rely on scene-level activity cues rather than reason about emerging harm.

We present AnyCap, a plug-and-play framework that brings instruction alignment to omni-modal captioning. Captions offer a unified language interface for multimodal learning, and users increasingly expect instruction-driven control over their content and style. Current caption models lack explicit instruction supervision and are weak at instruction following, while directly tuning them can degrade general language ability. Achieving instruction alignment in an omni-modal setting is harder still, as each modality calls for separate models and custom designs. To address these challenges, AnyCap leverages a multimodal revision-based refinement paradigm that refines uncontrolled captions from existing models to instruction-aligned ones, without re-training base models. By processing multi-modality features in a unified framework, it enables one model to serve images, videos, and audio. To address the lack of instruction-based data, we construct AnyCapData, a large-scale, high-quality corpus spanning three modalities with 28 well-designed instruction types, each with multiple instruction samples. For evaluation, we address the limitations of current metrics for instruction-oriented captioning by designing AnyCapEval. Its key insight is to decouple evaluation into content and style for fine-grained assessment. Extensive experiments show that on AnyCapEval and diverse public benchmarks, AnyCap consistently improves both caption quality and instruction adherence for both open-source and API-based models. Notably, AnyCap-8B boosts GPT-4o’s content scores by 46% and style scores by 12%. Our code and models will be made publicly available.

Competitive programming remains challenging for code LLMs. Despite recent progress, many training pipelines still depend on scarce real-world data, raising concerns about scalability and near-duplicate benchmark contamination. In this paper, we examine whether synthetic training artifacts can support the complete SFT-to-RL cycle for competitive programming: no real-world tasks, solutions, or test cases are directly used for post-training. To this end, we synthesize tasks, verified solutions, and reliable test cases that serve as reward signals for reinforcement learning. To improve reward reliability, we introduce a dual-verification strategy that reduces noise in both selected solutions and test outputs. Using this high-quality data, we train the X-Coder model series. X-Coder-14B achieves 67.5% avg@8 on LiveCodeBench v5 and 63.4% on v6, outperforming its base model by over 40 points. Further analysis provides practical insights into synthetic post-training, highlighting the value of diverse tasks, verified long-CoT supervision, and code-centric reinforcement. Our data and models are released at https://github.com/JieWu02/X-Coder.

We introduce SocialArena, a benchmark for systematically defining and evaluating the social intelligence of large language models (LLMs). Although LLMs are increasingly deployed in socially interactive applications such as mental health counseling, companionship chatbots, and educational tutoring, existing evaluations remain fragmented—focusing on isolated capabilities or single scenarios—and evaluation reliability is largely unverified. This work makes three main contributions to address these limitations. (1) We first propose a unified three-tier framework that defines social intelligence along three capability dimensions: role-playing ability, mental state reasoning, and behavioral reasoning. To operationalize this definition, we design a role-playing dialogue paradigm with explicit character thinking for evaluation, enabling simultaneous assessment of these capabilities in a manner that mirrors realistic human interactions. (2) To ensure diversity across social scenarios and thereby prevent evaluation conclusions from being confined to specific scenarios, we introduce the U-TRIMF framework, which guides us in synthesizing social contexts with broad coverage. (3) To obtain reliable evaluation, we invest 132.4K to recruit and train human annotators and adopt a human-annotated arena-style protocol, where annotators compare model responses and provide preference labels along with natural-language critiques. Experiments on a wide range of frontier LLMs reveal substantial gaps in their social intelligence capabilities. We hope that SocialArena can provide valuable insights and guidance for future LLM improvements.

Safety alignment is a fundamental prerequisite for building trustworthy artificial general intelligence. Despite substantial progress in safety alignment techniques, empirical evidence shows that aligned large language models can still produce unsafe responses under minor internal perturbations, revealing a robustness gap in existing safety mechanisms at the latent representation level. In this paper, we study the robustness evaluation of safety alignment under latent-space perturbations. We introduce Activation Steering Attack (ASA), and leverage the Negative Log-Likelihood (NLL) as a diagnostic signal to probe the local sensitivity of safety behaviors in latent space. By measuring a model’s likelihood under controlled perturbations to its hidden representations, we assess the stability of its original responses. The probing signal is model-agnostic and supervision-free, enabling a general and reproducible diagnostic metric for analyzing safety robustness. Leveraging these probes, we systematically uncover a set of previously underexplored empirical findings, including (1) non-stationarity of layer vulnerabilities, revealing that the most vulnerable layer is an unstable property and even relocates after robustness training; (2) instance-level alignment with cross-layer consistency, where specific inputs remain universally vulnerable across the entire model hierarchy; (3) compositional effects of ASA, characterized by its incremental accumulation across sequential decoding steps and its potential for prompt-level jailbreak effectiveness.

Evaluating Large Language Models (LLMs) has become increasingly important, with automatic evaluation benchmarks gaining prominence as alternatives to human evaluation. While existing research has focused on approximating model rankings, such benchmarks fail to provide users and developers with a comprehensive and fine-grained understanding of a specific model’s capabilities. To fill this gap, we propose SCAN (Structured Capability Assessment and Navigation), a practical framework that enables detailed characterization of LLM capabilities through comprehensive and fine-grained evaluation. SCAN incorporates four key components: (1) TaxBuilder, which extracts capability-indicating tags from extensive queries to construct a hierarchical taxonomy automatically; (2) RealMix, a query synthesis and filtering mechanism that ensures sufficient evaluation data for each capability tag; (3) a suite of visualization and analysis tools that facilitate efficient navigation and analysis of model capabilities; and (4) a PC²-based (Pre-Comparison-derived Criteria) LLM-as-a-Judge approach that achieves significantly higher accuracy compared to classic LLM-as-a-Judge method. Using SCAN, we conduct a comprehensive evaluation of 21 mainstream LLMs. Our detailed analysis of the GPT-OSS family reveals substantial performance variations, even within sub-capabilities belonging to the same category of capability. This finding highlights the importance of fine-grained evaluation in accurately understanding LLM behavior. Project homepage and resources are available at https://github.com/liudan193/SCAN.

Language model families exhibit striking disparity in their capacity to benefit from reinforcement learning: under identical training, models like Qwen achieve substantial gains, while others like Llama yield limited improvements. Complementing data-centric approaches, we reveal that this disparity reflects a hidden structural property: distributional clarity in probability space. Through a three-stage analysis—from phenomenon to mechanism to interpretation—we uncover that RL-friendly models exhibit intra-class compactness and inter-class separation in their probability assignments to correct vs. incorrect responses. We quantify this clarity using the Silhouette Coefficient (S) and demonstrate that (1) high S correlates strongly with RL performance; (2) low S is associated with severe logic errors and reasoning instability. To confirm this property, we introduce a Silhouette-Aware Reweighting strategy that prioritizes low-S samples during training. Experiments across six mathematical benchmarks show consistent improvements across all model families, with gains up to 5.9 points on AIME24. Our work establishes distributional clarity as a fundamental, trainable property underlying RL-Friendliness.

Large language models (LLMs) have exhibited impressive capabilities across a myriad of tasks, yet they occasionally yield undesirable outputs. We posit that these limitations are rooted in the foundational autoregressive architecture of LLMs, which inherently lacks mechanisms for differentiating between desirable and undesirable results. Drawing inspiration from the dual-process theory of human cognition, we introduce LLM2, a novel framework that combines an LLM (System 1) with a process-based verifier (System 2). Within LLM2, the LLM is responsible for generating plausible candidates, while the verifier provides timely process-based feedback to distinguish desirable and undesirable outputs. The verifier is trained with a pairwise comparison loss on synthetic process-supervision data generated through our token quality exploration strategy. Empirical results on mathematical reasoning benchmarks substantiate the efficacy of LLM2, exemplified by an accuracy enhancement from 50.3 to 57.8 (+7.5) for Llama3-1B on GSM8K. Furthermore, when combined with self-consistency, LLM2 achieves additional improvements, boosting major@20 accuracy from 56.2 to 70.2 (+14.0).

Text-Attributed Graphs (TAGs), which integrate text and graph structures, have recently gained traction, especially in web applications. However, as a graph structure, TAG representation learning (TAGRL) naturally inherits issues from Graph Neural Networks (GNNs), such as fairness. Moreover, previous TAGRL research has mainly focused on using LM-as-encoder to boost downstream task performance, with little consideration given to whether this process may raise additional concerns related to fairness and other safety-related issues. As the first work to explore fairness in TAGRL, this paper proposes the concept of evolving LM-as-encoder to LM-as-fair-encoder, developing a two-stage fairness-aware alignment process called FairTAG based on the observed issues. Specifically, we first mitigate the tendency of LMs to overfit to homophily during downstream tasks fine-tuning, followed by subgraph-level connection behavior preference optimization for selected anchor nodes. We provide theoretical support and demonstrate the feasibility of LM-as-fair-encoder through extensive experiments and ablation studies. We also show that FairTAG can be seamlessly integrated with fairness-enhancing strategies on the GNNs decoder side, thus innovatively constructing a plug-and-play learning framework.

Two-step approaches combining pre-trained large language model embeddings and anomaly detectors demonstrate strong performance in text anomaly detection by leveraging rich semantic representations. However, high-dimensional dense embeddings extracted by large language models pose challenges due to substantial memory requirements and high computation time. To address this challenge, we introduce the Simplified Isolation Kernel (SIK), which maps high-dimensional dense embeddings to lower-dimensional sparse representations while preserving crucial anomaly characteristics. SIK has linear-time complexity and significantly reduces space complexity through its innovative boundary-focused feature mapping.Experiments across 7 datasets demonstrate that SIK achieves better detection performance than 11 SOTA anomaly detection algorithms while maintaining computational efficiency and low memory cost. All code and demonstrations are available at https://github.com/charles-cao/SIK.

Large vision-language models (LVLMs) have shown remarkable performance in visual-language understanding for downstream multimodal tasks. While their capabilities are improving, problems emerge simultaneously. Among those problems, the hallucinations have attracted much attention, which stands for the phenomenon where LVLMs generate contradictory content to their input visual and text contents. Many approaches have been proposed to deal with this issue, such as contrastive decoding and attention manipulation. However, contrastive decoding methods struggle in constructing appropriate contrastive samples, and attention manipulation methods are highly sensitive, lacking stability. In this work, we propose image head Masked Contrastive Decoding (MaskCD). Our approach utilizes the “image heads” in LVLMs, masking them to construct contrastive samples for contrastive decoding. We evaluated MaskCD on LLaVA-1.5-7b and Qwen-VL-7b, using various benchmarks such as CHAIR, POPE, AMBER and MME. The results demonstrate that MaskCD effectively alleviates the phenomenon of hallucinations and retains the general capabilities of LVLMs. Corresponding resources could be found at: https://github.com/Deng-Jingyuan/MaskCD.

LLM Unlearning plays a crucial role in removing sensitive information from language models to mitigate potential misuse. However, previous approaches often treat nonsensical responses or template-based refusals (e.g., “Sorry, I cannot answer.”) as the unlearning target, which can give the impression of deliberate information suppression, making the process even more vulnerable to attacks and jailbreaks. Moreover, most methods rely on auxiliary models or retaining datasets, which adds complexity to the unlearning process. To address these challenges, we propose MEOW, a streamlined and stealthy unlearning method that eliminates the need for auxiliary models or retaining data while avoiding leakage through its innovative use of inverted facts. These inverted facts are generated by an offline LLM and serve as fine-tuning labels. Meanwhile, we introduce MEMO, a novel metric that measures the model’s memorization, to select optimal fine-tuning targets. The use of inverted facts not only maintains the covert nature of the model but also ensures that sensitive information is effectively forgotten without revealing the target data. Evaluated on the ToFU Knowledge Unlearning dataset using Llama2-7B-Chat and Phi-1.5, MEOW outperforms baselines in forgetting quality while preserving model utility. MEOW also maintains strong performance across NLU and NLG tasks and demonstrates superior resilience to attacks, validated via the Min-K% membership inference method.

Embeddings as a Service (EaaS) is emerging as a crucial role in AI applications. Unfortunately, EaaS is vulnerable to model extraction attacks, highlighting the urgent need for copyright protection. Although some preliminary works propose applying embedding watermarks to protect EaaS, recent research reveals that these watermarks can be easily removed. Hence, it is crucial to inject robust watermarks resistant to watermark removal attacks. Existing watermarking methods typically inject a target embedding into embeddings through linear interpolation when the text contains triggers. However, this mechanism results in each watermarked embedding having the same component, which makes the watermark easy to identify and eliminate. Motivated by this, in this paper, we propose a novel embedding-specific watermarking (ESpeW) mechanism to offer robust copyright protection for EaaS. Our approach involves injecting unique, yet readily identifiable watermarks into each embedding. Watermarks inserted by ESpeW are designed to maintain a significant distance from one another and to avoid sharing common components, thus making it significantly more challenging to remove the watermarks. Moreover, ESpeW is minimally invasive, as it reduces the impact on embeddings to less than 1%, setting a new milestone in watermarking for EaaS. Extensive experiments on four popular datasets demonstrate that ESpeW can even watermark successfully against a highly aggressive removal strategy without sacrificing the quality of embeddings.

Large Language Models (LLMs), constrained by limited context windows, often face significant performance degradation when reasoning over long contexts. To address this, Retrieval-Augmented Generation (RAG) retrieves and reasons over chunks but frequently sacrifices logical coherence due to its reliance on similarity-based rankings. Similarly, divide-and-conquer frameworks (DCF) split documents into small chunks for independent reasoning and aggregation. While effective for local reasoning, DCF struggles to capture long-range dependencies and risks inducing conflicts by processing chunks in isolation. To overcome these limitations, we propose ToM, a novel Tree-oriented MapReduce framework for long-context reasoning. ToM leverages the inherent hierarchical structure of long documents (e.g., main headings and subheadings) by constructing a DocTree through hierarchical semantic parsing and performing bottom-up aggregation. Using a Tree MapReduce approach, ToM enables recursive reasoning: in the Map step, rationales are generated at child nodes; in the Reduce step, these rationales are aggregated across sibling nodes to resolve conflicts or reach consensus at parent nodes. Experimental results on 70B+ LLMs show that ToM significantly outperforms existing divide-and-conquer frameworks and retrieval-augmented generation methods, achieving better logical coherence and long-context reasoning.

Preference learning extends the performance of Code LLMs beyond traditional supervised fine-tuning by leveraging relative quality comparisons. In existing approaches, a set of n candidate solutions is evaluated based on test case success rates, with the candidate demonstrating a higher pass rate being labeled as positive and its counterpart with a lower pass rate as negative. However, because this approach aligns entire failing code blocks rather than pinpointing specific errors, it lacks the granularity necessary to capture meaningful error-correction relationships. As a result, the model is unable to learn more informative error-correction patterns. To address these issues, we propose Target-DPO, a new preference alignment framework that mimics human iterative debugging to refine Code LLMs. Target-DPO explicitly locates error regions and aligns the corresponding tokens via a tailored DPO algorithm. To facilitate it, we introduce the CodeFlow dataset, where samples are iteratively refined until passing tests, with modifications capturing error corrections. Extensive experiments show that a diverse suite of Code LLMs equipped with Target-DPO achieves significant performance gains in code generation and improves on challenging tasks like BigCodeBench. In-depth analysis reveals that Target-DPO yields fewer errors. Code, model and datasets are in: https://github.com/JieWu02/Target-DPO.

Logit-based LLM watermarking traces and verifies AI-generated content by maintaining green and red token lists and increasing the likelihood of green tokens during generation. However, it struggles in low-entropy scenarios, where predictable outputs make green token selection difficult without disrupting natural text flow. Existing approaches address this by assuming access to the original LLM to calculate entropy and selectively watermark high-entropy tokens. However, these methods face two major challenges: (1) high computational costs and detection delays due to reliance on the original LLM, and (2) potential risks of model leakage. To address these limitations, we propose Invisible Entropy (IE), a watermarking paradigm designed to enhance both safety and efficiency. Instead of relying on the original LLM, IE introduces a lightweight feature extractor and an entropy tagger to predict whether the entropy of the next token is high or low. Furthermore, based on theoretical analysis, we developed a threshold navigator that adaptively sets entropy thresholds. It identifies a threshold where the watermark ratio decreases as the green token count increases, enhancing the naturalness of the watermarked text and improving detection robustness. Experiments on HumanEval and MBPP datasets demonstrate that IE reduces parameter size by 99% while achieving performance on par with state-of-the-art methods: https://anonymous.4open.science/r/IE-Official.

Distilling high-accuracy Graph Neural Networks (GNNs) to low-latency multilayer perceptrons (MLPs) on graph tasks has become a hot research topic. However, conventional MLP learning relies almost exclusively on graph nodes and fails to effectively capture the graph structural information. Previous methods address this issue by processing graph edges into extra inputs for MLPs, but such graph structures may be unavailable for various scenarios. To this end, we propose Prototype-Guided Knowledge Distillation (PGKD), which does not require graph edges (edge-free setting) yet learns structure-aware MLPs. Our insight is to distill graph structural information from GNNs. Specifically, we first employ the class prototypes to analyze the impact of graph structures on GNN teachers, and then design two losses to distill such information from GNNs to MLPs. Experimental results on popular graph benchmarks demonstrate the effectiveness and robustness of the proposed PGKD.

Watermarking by altering token sampling probabilities based on red-green list is a promising method for tracing the origin of text generated by large language models (LLMs). However, existing watermark methods often struggle with a fundamental dilemma: improving watermark effectiveness (the detectability of the watermark) often comes at the cost of reduced text quality. This trade-off limits their practical application. To address this challenge, we first formalize the problem within a multi-objective trade-off analysis framework. Within this framework, we identify a key factor that influences the dilemma. Unlike existing methods, where watermark strength is typically treated as a fixed hyperparameter, our theoretical insights lead to the development of MorphMark—a method that adaptively adjusts the watermark strength in response to changes in the identified factor, thereby achieving an effective resolution of the dilemma. In addition, MorphMark also prioritizes flexibility since it is an model-agnostic and model-free watermark method, thereby offering a practical solution for real-world deployment, particularly in light of the rapid evolution of AI models. Extensive experiments demonstrate that MorphMark achieves a superior resolution of the effectiveness-quality dilemma, while also offering greater flexibility and time and space efficiency.

Although fine-tuning Large Language Models (LLMs) with multilingual data can rapidly enhance the multilingual capabilities of LLMs, they still exhibit a performance gap between the dominant language (e.g., English) and non-dominant ones due to the imbalance of training data across languages. To further enhance the performance of non-dominant languages, we propose ShifCon, a Shift-based multilingual Contrastive framework that aligns the internal forward process of other languages toward that of the dominant one. Specifically, it shifts the representations of non-dominant languages into the dominant language subspace, allowing them to access relatively rich information encoded in the model parameters. The enriched representations are then shifted back into their original language subspace before generation. Moreover, we introduce a subspace distance metric to pinpoint the optimal layer area for shifting representations and employ multilingual contrastive learning to further enhance the alignment of representations within this area. Experiments demonstrate that our ShifCon framework significantly enhances the performance of non-dominant languages, particularly for low-resource ones. Further analysis offers extra insights to verify the effectiveness of ShifCon and propel future research.

Large Language Models (LLMs) have made notable progress in mathematical reasoning, yet they often rely on single-paradigm reasoning that limits their effectiveness across diverse tasks. In this paper, we introduce Chain-of-Reasoning (CoR), a novel unified framework that integrates multiple reasoning paradigms — Natural Language Reasoning (NLR), Algorithmic Reasoning (AR), and Symbolic Reasoning (SR) — to enable synergistic collaboration. CoR generates multiple potential answers using different reasoning paradigms and synthesizes them into a coherent final solution. We propose a Progressive Paradigm Training (PPT) strategy that allows models to progressively master these paradigms, culminating in the development of at CoR-Math-7B. Experimental results demonstrate that CoR-Math-7B significantly outperforms current SOTA models, achieving up to a 41.0% absolute improvement over GPT-4o in theorem proving tasks and a 15% improvement over RL-based methods on the MATH benchmark in arithmetic tasks. These results show the enhanced mathematical comprehensive ability of our model, enabling zero-shot generalization across tasks.The code is available at https://github.com/microsoft/CoR.

Word-level AutoCompletion (WLAC) is a rewarding yet challenging task in Computer-aided Translation. Existing work addresses this task through a classification model based on a neural network that maps the hidden vector of the input context into its corresponding label (i.e., the candidate target word is treated as a label). Since the context hidden vector itself does not take the label into account and it is projected to the label through a linear classifier, the model cannot sufficiently leverage valuable information from the source sentence as verified in our experiments, which eventually hinders its overall performance. To alleviate this issue, this work proposes an energy-based model for WLAC, which enables the context hidden vector to capture crucial information from the source sentence. Unfortunately, training and inference suffer from efficiency and effectiveness challenges, therefore we employ three simple yet effective strategies to put our model into practice. Experiments on four standard benchmarks demonstrate that our reranking-based approach achieves substantial improvements (about 6.07%) over the previous state-of-the-art model. Further analyses show that each strategy of our approach contributes to the final performance.1

Large language models (LLMs) have demonstrated impressive capabilities in general scenarios, exhibiting a level of aptitude that approaches, in some aspects even surpasses, human-level intelligence. Among their numerous skills, the translation abilities of LLMs have received considerable attention. Compared to typical machine translation that focuses solely on source-to-target mapping, LLM-based translation can potentially mimic the human translation process, which might take preparatory steps to ensure high-quality translation. This work explores this possibility by proposing the MAPS framework, which stands for Multi-Aspect Prompting and Selection. Specifically, we enable LLMs first to analyze the given source sentence and induce three aspects of translation-related knowledge (keywords, topics, and relevant demonstrations) to guide the final translation process. Moreover, we employ a selection mechanism based on quality estimation to filter out noisy and unhelpful knowledge. Both automatic (3 LLMs × 11 directions × 2 automatic metrics) and human evaluation (preference study and MQM) demonstrate the effectiveness of MAPS. Further analysis shows that by mimicking the human translation process, MAPS reduces various translation errors such as hallucination, ambiguity, mistranslation, awkward style, untranslated text, and omission. Source code is available at https://github.com/zwhe99/MAPS-mt.

Instruction tuning effectively optimizes Large Language Models (LLMs) for downstream tasks. Due to the changing environment in real-life applications, LLMs necessitate continual task-specific adaptation without catastrophic forgetting. Considering the heavy computational cost, replay-based Continual Learning (CL) methods are the simplest and most widely used for LLMs to address the forgetting issue. However, traditional replay-based methods do not fully utilize instructions to customize the replay strategy. In this work, we propose a novel paradigm called Instruction-based Continual Learning (InsCL). InsCL dynamically replays previous data based on task similarity, calculated by Wasserstein Distance with instructions. Moreover, we further introduce an Instruction Information Metric (InsInfo) to quantify the complexity and diversity of instructions. According to InsInfo, InsCL guides the replay process more inclined to high-quality data. We conduct extensive experiments over 16 tasks with different training orders, observing consistent performance improvements of InsCL. When all tasks have been trained, InsCL achieves performance gains of 3.0 Relative Gain compared with Random Replay, and 27.96 Relative Gain compared with No Replay.

Knowledge Graph Completion (KGC) is a task aimed at uncovering the inherent relationships among known knowledge triplets in a Knowledge Graph (KG) and subsequently predicting missing links. Presently, there is a rising interest in inductive knowledge graph completion, where missing links may pertain to previously unobserved entities. Previous inductive KGC methods mainly rely on descriptive information of entities to improve the representation of unseen entities, neglecting to provide effective prior knowledge for relation modeling. To tackle this challenge, we capture prior schema-level interactions related to relations by leveraging entity type information, thereby furnishing effective prior constraints when reasoning with newly introduced entities. Moreover, We employ normal in-batch negatives and introduce schema-guided negatives to bolster the efficiency of normal contrastive representation learning. Experimental results demonstrate that our approach consistently achieves state-of-the-art performance on various established metrics across multiple benchmark datasets for link prediction. Notably, our method achieves a 20.5% relative increase in Hits@1 on the HumanWiki-Ind dataset.

Knowledge Distillation (KD) is a predominant approach for BERT compression.Previous KD-based methods focus on designing extra alignment losses for the student model to mimic the behavior of the teacher model.These methods transfer the knowledge in an indirect way.In this paper, we propose a novel Weight-Inherited Distillation (WID), which directly transfers knowledge from the teacher.WID does not require any additional alignment loss and trains a compact student by inheriting the weights, showing a new perspective of knowledge distillation.Specifically, we design the row compactors and column compactors as mappings and then compress the weights via structural re-parameterization.Experimental results on the GLUE and SQuAD benchmarks show that WID outperforms previous state-of-the-art KD-based baselines.Further analysis indicates that WID can also learn the attention patterns from the teacher model without any alignment loss on attention distributions.The code is available at https://github.com/wutaiqiang/WID-NAACL2024.

Generative AI has demonstrated unprecedented creativity in the field of computer vision, yet such phenomena have not been observed in natural language processing. In particular, large language models (LLMs) can hardly produce written works at the level of human experts due to the extremely high complexity of literature writing. In this paper, we present HoLLMwood, an automated framework for unleashing the creativity of LLMs and exploring their potential in screenwriting, which is a highly demanding task. Mimicking the human creative process, we assign LLMs to different roles involved in the real-world scenario. In addition to the common practice of treating LLMs as Writer, we also apply LLMs as Editor, who is responsible for providing feedback and revision advice to Writer. Besides, to enrich the characters and deepen the plots, we introduce a role-playing mechanism and adopt LLMs as Actors that can communicate and interact with each other. Evaluations on automatically generated screenplays show that HoLLMwood substantially outperforms strong baselines in terms of coherence, relevance, interestingness and overall quality.

The ability of Large Language Models (LLMs) to critique and refine their reasoning is crucial for their application in evaluation, feedback provision, and self-improvement. This paper introduces CriticBench, a comprehensive benchmark designed to assess LLMs’ abilities to critique and rectify their reasoning across a variety of tasks. CriticBench encompasses five reasoning domains: mathematical, commonsense, symbolic, coding, and algorithmic. It compiles 15 datasets and incorporates responses from three LLM families. Utilizing CriticBench, we evaluate and dissect the performance of 17 LLMs in generation, critique, and correction reasoning, i.e., GQC reasoning. Our findings reveal: (1) a linear relationship in GQC capabilities, with critique-focused training markedly enhancing performance; (2) a task-dependent variation in correction effectiveness, with logic-oriented tasks being more amenable to correction; (3) GQC knowledge inconsistencies that decrease as model size increases; and (4) an intriguing inter-model critiquing dynamic, where stronger models are better at critiquing weaker ones, while weaker models can surprisingly surpass stronger ones in their self-critique. We hope these insights into the nuanced critique-correct reasoning of LLMs will foster further research in LLM critique and self-improvement.

Neural machine translation (NMT) systems often produce inadequate translations for named entities. In this study, we conducted preliminary experiments to examine the factors affecting the translation accuracy of named entities, specifically focusing on their translation difficulty and context diversity. Based on our observations, we propose a novel data augmentation strategy to enhance the accuracy of named entity translation. The main concept behind our approach is to increase both the context diversity and translation probability for the targeted named entity pair. To achieve this, we construct additional samples for named entities that exhibit high translation difficulty or low context diversity and use the augmented training data to re-train the final translation model. Furthermore, we propose an entity-aware machine translation metric that prefers the translation output to generate more accurate named entities. Our experimental results demonstrate significant improvements over the baseline in terms of general translation performance and named entity translation accuracy across various test sets, such as WMT news translation and terminology test sets.

While large language models (LLMs) have achieved impressive performance across diverse tasks, recent studies showcase that causal LLMs suffer from the “reversal curse”. It is a typical example that the model knows “A’s father is B”, but is unable to reason “B’s child is A”. This limitation poses a challenge to the advancement of artificial general intelligence (AGI), as it suggests a gap in the models’ ability to comprehend and apply bidirectional reasoning. In this paper, we first conduct substantial evaluation and identify that the root cause of the reversal curse lies in the different word order between the training and inference stage, namely, the poor ability of causal language models to predict antecedent words within the training data. Accordingly, permutation on the training data is considered as a potential solution, since this can make the model predict antecedent words or tokens. However, previous permutation methods may disrupt complete phrases or entities, thereby posing challenges for the model to comprehend and learn from training data. To address this issue, we propose Semantic-aware Permutation Training (SPT), which addresses this issue by segmenting the training sentences into semantic units (i.e., entities or phrases) with an assistant language model and permuting these units before feeding into the model. Extensive experiments demonstrate that SPT effectively mitigates the reversal curse since the performance on reversed questions approximates that on the forward ones, and significantly advances the performance of existing works.

Large language models with instruction-following abilities have revolutionized the field of artificial intelligence. These models show exceptional generalizability to tackle various real-world tasks through their natural language interfaces. However, their performance heavily relies on high-quality exemplar data, which is often difficult to obtain. This challenge is further exacerbated when it comes to multimodal instruction following. We introduce TextBind, an almost annotation-free framework for empowering LLMs with multi-turn interleaved multimodal instruction-following capabilities. Our approach requires only image-caption pairs and generates multi-turn multimodal instruction-response conversations from a language model. To accommodate interleaved image-text inputs and outputs, we devise MIM, a language model-centric architecture that seamlessly integrates image encoder and decoder models. Extensive quantitative and qualitative experiments demonstrate that MIM trained on TextBind achieves remarkable generation capability in multimodal conversations compared to recent baselines.

Modern large language models (LLMs) like ChatGPT have shown remarkable performance on general language tasks but still struggle on complex reasoning tasks, which drives the research on cognitive behaviors of LLMs to explore human-like problem-solving strategies. Along this direction, one representative strategy is self-reflection, which asks an LLM to refine the solution with the feedback generated by itself iteratively. However, our study shows that such reflection-style methods suffer from the Degeneration-of-Thought (DoT) problem: once the LLM has established confidence in its solutions, it is unable to generate novel thoughts later through reflection even if its initial stance is incorrect. To address the DoT problem, we propose a Multi-Agent Debate (MAD) framework, in which multiple agents express their arguments in the state of “tit for tat” and a judge manages the debate process to obtain a final solution. Clearly, our MAD framework encourages divergent thinking in LLMs which would be helpful for tasks that require deep levels of contemplation. Experiment results on two challenging datasets, commonsense machine translation and counter-intuitive arithmetic reasoning, demonstrate the effectiveness of our MAD framework. Extensive analyses suggest that the adaptive break of debate and the modest level of “tit for tat” state are required for MAD to obtain good performance. Moreover, we find that LLMs might not be a fair judge if different LLMs are used for agents.

Scientific reasoning poses an excessive challenge for even the most advanced Large Language Models (LLMs). To make this task more practical and solvable for LLMs, we introduce a new task setting named tool-augmented scientific reasoning. This setting supplements LLMs with scalable toolsets, and shifts the focus from pursuing an omniscient problem solver to a proficient tool-user. To facilitate the research of such setting, we construct a tool-augmented training corpus named MathFunc which encompasses over 30,000 samples and roughly 6,000 tools. Building on MathFunc, we develop SciAgent to retrieve, understand and, if necessary, use tools for scientific problem solving. Additionally, we craft a benchmark, SciToolBench, spanning five scientific domains to evaluate LLMs’ abilities with tool assistance. Extensive experiments on SciToolBench confirm the effectiveness of SciAgent. Notably, SciAgent-Llama3-8B surpasses other LLMs with the comparable size by more than 8.0% in absolute accuracy. Furthermore, SciAgent-DeepMath-7B shows much superior performance than ChatGPT.

Tool-augmented large language models (LLMs) are rapidly being integrated into real-world applications. Due to the lack of benchmarks, the community has yet to fully understand the hallucination issues within these models. To address this challenge, we introduce a comprehensive diagnostic benchmark, ToolBH. Specifically, we assess the LLM’s hallucinations through two perspectives: depth and breadth. In terms of depth, we propose a multi-level diagnostic process, including (1) solvability detection, (2) solution planning, and (3) missing-tool analysis. For breadth, we consider three scenarios based on the characteristics of the toolset: missing necessary tools, potential tools, and limited functionality tools. Furthermore, we developed seven tasks and collected 700 evaluation samples through multiple rounds of manual annotation. The results show the significant challenges presented by the ToolBH benchmark. The current advanced models Gemini-1.5-Pro and GPT-4o only achieve total scores of 45.3 and 37.0, respectively, on a scale of 100. In this benchmark, larger model parameters do not guarantee better performance; the training data and response strategies also play crucial roles in tool-enhanced LLM scenarios. Our diagnostic analysis indicates that the primary reason for model errors lies in assessing task solvability. Additionally, open-weight models suffer from performance drops with verbose replies, whereas proprietary models excel with longer reasoning.

Decoding methods play an indispensable role in converting language models from next-token predictors into practical task solvers. Prior research on decoding methods, primarily focusing on task-specific models, may not extend to the current era of general-purpose large language models (LLMs). Moreover, the recent influx of decoding strategies has further complicated this landscape. This paper provides a comprehensive and multifaceted analysis of various decoding methods within the context of LLMs, evaluating their performance, robustness to hyperparameter changes, and decoding speeds across a wide range of tasks, models, and deployment environments. Our findings reveal that decoding method performance is notably task-dependent and influenced by factors such as alignment, model size, and quantization. Intriguingly, sensitivity analysis exposes that certain methods achieve superior performance at the cost of extensive hyperparameter tuning, highlighting the trade-off between attaining optimal results and the practicality of implementation in varying contexts.

Large Language Models (LLMs) have emerged as powerful tools for Text-to-SQL tasks, exhibiting remarkable reasoning capabilities. Different from tasks such as math word problem and commonsense reasoning, SQL solutions have a relatively fixed pattern. This facilitates the investigation of whether LLMs can benefit from categorical thinking, mirroring how humans acquire knowledge through inductive reasoning based on comparable examples. In this study, we propose that employing query group partitioning allows LLMs to focus on learning the thought processes specific to a single problem type, consequently enhancing their reasoning abilities across diverse difficulty levels and problem categories. Our experiments reveal that multiple advanced LLMs, when equipped with PTD-SQL, can either surpass or match previous state-of-the-art (SOTA) methods on the Spider and BIRD datasets. Intriguingly, models with varying initial performances have exhibited significant improvements mainly at the boundary of their capabilities after targeted drilling, suggesting a parallel with human progress. Code is available at https://github.com/lrlbbzl/PTD-SQL.

In the task of incremental few-shot relation classification, model performance is always limited by the incompatibility between the base feature embedding space and the novel feature embedding space. To tackle the issue, we propose a novel model named ICA-Proto: Iterative Cross Alignment prototypical network. Specifically, we incorporate the query representation into the encoding of novel prototypes and utilize the query-aware prototypes to update the query representation at the same time. Further, we implement the above process iteratively to achieve more interaction. In addition, a novel prototype quadruplet loss is designed to regulate the spatial distributions of embedding space, so as to make it easier for the relation classification. Experimental results on two benchmark datasets demonstrate that ICA-Proto significantly outperforms the state-of-the-art baseline model.

Human conversations contain natural and reasonable topic shifts, reflected as the concept flows across utterances. Previous researches prove that explicitly modeling concept flows with a large commonsense knowledge graph effectively improves response quality. However, we argue that there exists a gap between the knowledge graph and the conversation. The knowledge graph has limited commonsense knowledge and ignores the characteristics of natural conversations. Thus, many concepts and relations in conversations are not included. To bridge this gap, we propose to enhance dialogue generation with conversational concept flows. Specifically, we extract abundant concepts and relations from natural conversations and build a new conversation-aware knowledge graph. In addition, we design a novel relation-aware graph encoder to capture the concept flows guided by the knowledge graph. Experimental results on the large-scale Reddit conversation dataset indicate that our method performs better than strong baselines, andfurther analysis verifies the effectiveness of each component. All our code and data will be publicly available after acceptance.

Hot news is one of the most popular topics in daily conversations. However, news grounded conversation has long been stymied by the lack of well-designed task definition and scarce data. In this paper, we propose a novel task, Proactive News Grounded Conversation, in which a dialogue system can proactively lead the conversation based on some key topics of the news. In addition, both information-seeking and chit-chat scenarios are included realistically, where the user may ask a series of questions about the news details or express their opinions and be eager to chat. To further develop this novel task, we collect a human-to-human Chinese dialogue dataset NewsDialogues, which includes 1K conversations with a total of 14.6K utterances and detailed annotations for target topics and knowledge spans. Furthermore, we propose a method named Predict-Generate-Rank, consisting of a generator for grounded knowledge prediction and response generation, and a ranker for the ranking of multiple responses to alleviate the exposure bias. We conduct comprehensive experiments to demonstrate the effectiveness of the proposed method and further present several key findings and challenges to prompt future research.

The potential of large language models (LLMs) to simultaneously perform a wide range of natural language processing (NLP) tasks has been the subject of extensive research. Although instruction tuning has proven to be a data-efficient method for transforming LLMs into such generalist models, their performance still lags behind specialist models trained exclusively for specific tasks. In this paper, we investigate whether incorporating broadcoverage generalist instruction tuning can contribute to building a specialist model. We hypothesize that its efficacy depends on task specificity and skill requirements. Our experiments assess four target tasks with distinct coverage levels, revealing that integrating generalist instruction tuning consistently enhances model performance when the task coverage is broad. The effect is particularly pronounced when the amount of task-specific training data is limited. Further investigation into three target tasks focusing on different capabilities demonstrates that generalist instruction tuning improves understanding and reasoning abilities. However, for tasks requiring factual knowledge, generalist data containing hallucinatory information may negatively affect the model’s performance. Overall, our work provides a systematic guide for developing specialist models with general instruction tuning.

Question answering plays a pivotal role in human daily life because it involves our acquisition of knowledge about the world. However, due to the dynamic and ever-changing nature of real-world facts, the answer can be completely different when the time constraint in the question changes. Recently, Large Language Models (LLMs) have shown remarkable intelligence in question answering, while our experiments reveal that the aforementioned problems still pose a significant challenge to existing LLMs. This can be attributed to the LLMs’ inability to perform rigorous reasoning based on surface-level text semantics. To overcome this limitation, rather than requiring LLMs to directly answer the question, we propose a novel approach where we reframe the Question Answering task as Programming (QAaP). Concretely, by leveraging modern LLMs’ superior capability in understanding both natural language and programming language, we endeavor to harness LLMs to represent diversely expressed text as well-structured code and select the best matching answer from multiple candidates through programming. We evaluate our QAaP framework on several time-sensitive question answering datasets and achieve decent improvement, up to 14.5% over strong baselines.

Information-seeking conversation, which aims to help users gather information through conversation, has achieved great progress in recent years. However, the research is still stymied by the scarcity of training data. To alleviate this problem, we propose AutoConv for synthetic conversation generation, which takes advantage of the few-shot learning ability and generation capacity of large language models (LLM). Specifically, we formulate the conversation generation problem as a language modeling task, then finetune an LLM with a few human conversations to capture the characteristics of the information-seeking process and use it for generating synthetic conversations with high quality. Experimental results on two frequently-used datasets verify that AutoConv has substantial improvements over strong baselines and alleviates the dependence on human annotation. In addition, we also provide several analysis studies to promote future research.

Large-scale pre-trained language models (PLMs) bring new opportunities to challenging problems, especially those that need high-level intelligence, such as the math word problem (MWPs). However, directly applying existing PLMs to MWPs can fail as the generation process lacks sufficient supervision and thus lacks fast adaptivity as humans. We notice that human reasoning has a dual reasoning framework that consists of an immediate reaction system (system 1) and a delicate reasoning system (system 2), where the entire reasoning is determined by their interaction. This inspires us to develop a cooperative reasoning-induced PLM for solving MWPs, called Cooperative Reasoning (CoRe), resulting in a human-like reasoning architecture with system 1 as the generator and system 2 as the verifier. In our approach, the generator is responsible for generating reasoning paths, and the verifiers are used to supervise the evaluation in order to obtain reliable feedback for the generator. We evaluate our CoRe framework on several mathematical reasoning datasets and achieve decent improvement over state-of-the-art methods, up to 9.6% increase over best baselines.

Generative methods greatly promote aspect-based sentiment analysis via generating a sequence of sentiment elements in a specified format. However, existing studies usually predict sentiment elements in a fixed order, which ignores the effect of the interdependence of the elements in a sentiment tuple and the diversity of language expression on the results. In this work, we propose Multi-view Prompting (MVP) that aggregates sentiment elements generated in different orders, leveraging the intuition of human-like problem-solving processes from different views. Specifically, MVP introduces element order prompts to guide the language model to generate multiple sentiment tuples, each with a different element order, and then selects the most reasonable tuples by voting. MVP can naturally model multi-view and multi-task as permutations and combinations of elements, respectively, outperforming previous task-specific designed methods on multiple ABSA tasks with a single model. Extensive experiments show that MVP significantly advances the state-of-the-art performance on 10 datasets of 4 benchmark tasks, and performs quite effectively in low-resource settings. Detailed evaluation verified the effectiveness, flexibility, and cross-task transferability of MVP.

This paper presents IIGroup’s submission to the WMT22 Word-Level AutoCompletion(WLAC) Shared Task in four language directions. We propose to use a Generate-then-Rerank framework to solve this task. More specifically, the generator is used to generate candidate words and recall as many positive candidates as possible. To facilitate the training process of the generator, we propose a span-level mask prediction task. Once we get the candidate words, we take the top-K candidates and feed them into the reranker. The reranker is used to select the most confident candidate. The experimental results in four language directions demonstrate the effectiveness of our systems. Our systems achieve competitive performance ranking 1st in English to Chinese subtask and 2nd in Chinese to English subtask.

In empathetic conversations, humans express their empathy to others with empathetic intents. However, most existing empathetic conversational methods suffer from a lack of empathetic intents, which leads to monotonous empathy. To address the bias of the empathetic intents distribution between empathetic dialogue models and humans, we propose a novel model to generate empathetic responses with human-consistent empathetic intents, EmpHi for short. Precisely, EmpHi learns the distribution of potential empathetic intents with a discrete latent variable, then combines both implicit and explicit intent representation to generate responses with various empathetic intents. Experiments show that EmpHi outperforms state-of-the-art models in terms of empathy, relevance, and diversity on both automatic and human evaluation. Moreover, the case studies demonstrate the high interpretability and outstanding performance of our model.

In Visual Question Answering (VQA), existing bilinear methods focus on the interaction between images and questions. As a result, the answers are either spliced into the questions or utilized as labels only for classification. On the other hand, trilinear models such as the CTI model efficiently utilize the inter-modality information between answers, questions, and images, while ignoring intra-modality information. Inspired by this observation, we propose a new trilinear interaction framework called MIRTT (Learning Multimodal Interaction Representations from Trilinear Transformers), incorporating the attention mechanisms for capturing inter-modality and intra-modality relationships. Moreover, we design a two-stage workflow where a bilinear model reduces the free-form, open-ended VQA problem into a multiple-choice VQA problem. Furthermore, to obtain accurate and generic multimodal representations, we pre-train MIRTT with masked language prediction. Our method achieves state-of-the-art performance on the Visual7W Telling task and VQA-1.0 Multiple Choice task and outperforms bilinear baselines on the VQA-2.0, TDIUC and GQA datasets.

With the transformation of education from the traditional classroom environment to online education and assessment, it is more and more important to accurately assess the difficulty of questions than ever. As teachers may not be able to follow the student’s performance and learning behavior closely, a well-defined method to measure the difficulty of questions to guide learning is necessary. In this paper, we explore the concept of question difficulty and provide our new Chinese DT-QDC dataset. This is currently the largest and only Chinese question dataset, and it also has enriched attributes and difficulty labels. Additional attributes such as keywords, chapter, and question type would allow models to understand questions more precisely. We proposed the MTMS-BERT and ORMS-BERT, which can improve the judgment of difficulty from different views. The proposed methods outperforms different baselines by 7.79% on F1-score and 15.92% on MAE, 28.26% on MSE on the new DT-QDC dataset, laying the foundation for the question difficulty comprehension task.

Recently deep learning has been used in Medical subject headings (MeSH) indexing to reduce the time and monetary cost by manual annotation, including DeepMeSH, TextCNN, etc. However, these models still suffer from failing to capture the complex correlations between MeSH terms. To this end, we introduce Graph Convolution Network (GCN) to learn the relationship between these terms, and present a novel Hybrid Graph Convolution Net for MeSH index (HGCN4MeSH). Basically, we utilize two BiGRUs to learn the embedding representation of the abstract and the title of the MeSH index text respectively. At the same time, we establish the adjacency matrix of MeSH terms based on the co-occurrence relationships in Corpus, which is easy to apply for GCN representation learning. On the basis of learning the mixed representation, the prediction problem of the MeSH index keywords is transformed into an extreme multi-label classification problem after the attention layer operation. Experimental results on two datasets show that HGCN4MeSH is competitive compared with the state-of-the-art methods.

In recent years, a variety of recurrent neural networks have been proposed, e.g LSTM. However, existing models only read the text once, it cannot describe the situation of repeated reading in reading comprehension. In fact, when reading or analyzing a text, we may read the text several times rather than once if we couldn’t well understand it. So, how to model this kind of the reading behavior? To address the issue, we propose a multi-glance mechanism (MGM) for modeling the habit of reading behavior. In the proposed framework, the actual reading process can be fully simulated, and then the obtained information can be consistent with the task. Based on the multi-glance mechanism, we design two types of recurrent neural network models for repeated reading: Glance Cell Model (GCM) and Glance Gate Model (GGM). Visualization analysis of the GCM and the GGM demonstrates the effectiveness of multi-glance mechanisms. Experiments results on the large-scale datasets show that the proposed methods can achieve better performance.

Deep learning approaches for sentiment classification do not fully exploit sentiment linguistic knowledge. In this paper, we propose a Multi-sentiment-resource Enhanced Attention Network (MEAN) to alleviate the problem by integrating three kinds of sentiment linguistic knowledge (e.g., sentiment lexicon, negation words, intensity words) into the deep neural network via attention mechanisms. By using various types of sentiment resources, MEAN utilizes sentiment-relevant information from different representation sub-spaces, which makes it more effective to capture the overall semantics of the sentiment, negation and intensity words for sentiment prediction. The experimental results demonstrate that MEAN has robust superiority over strong competitors.