Watching Synthetic Videos: Aligning Cross-modal Representations with Visual Synthesis for Zero-shot Video Captioning

Text-only training is a popular paradigm in zero-shot video captioning, where the video distribution is not available to the model during training, leading to a cross-modal gap between the training (text-only) and the inference (video-only). Previous works attempt to bridge the gap through simple linear transformations. However, the inherent gap between text and video makes cross-modal representation space alignment insufficient, resulting in inaccurate sentences. To address this issue, we propose a novel zero-shot video captioning framework (WSV) consisting of two training stages, which first generates corresponding synthetic video latent representations via a pretrained text-to-video generation model. To strengthen the fidelity of the latent representations, we propose a polisher capable of bridging the gap between real and synthetic video distributions. Subsequently, we design a prompter that conditions GPT-2 on the polished latent representations to generate the captions in the second training stage. During inference, an input video is encoded by a pretrained 3D Causal VAE and then fed directly into the prompter, which in turn guides GPT-2 to produce the final caption. Experimental results conducted on MSVD, MSR-VTT, and VATEX datasets demonstrate that our proposed method achieves scores of 52 and 95.7 on the B@4 and CIDEr metrics, respectively.

Adaptive Emotional Video Captioning via Affective Heterogeneous Graph Reasoning and Multi-task Joint Learning

Emotional video captioning (EVC) aims to describe a video with both factual correctness and affective expressiveness. It requires a model to perceive subtle, ambiguous, and temporally varying emotional cues and translate them into natural language without weakening objective visual content. Existing methods have progressively introduced contextual attention, emotion interpretation, emotion priors, dynamic emotion perception and emotion-cause reasoning. Nevertheless, most of them still depend on either global emotion vectors or rigid hierarchical priors. In recent methods, the tree-structured emotion prior establishes a coarse-to-fine connection between psychological emotion categories and daily emotion words, but its hard subordinate masking may irreversibly suppress correct lexical emotions once the coarse category prediction is inaccurate. It is also limited in representing mixed or overlapping emotions that frequently occur in real videos. To address the issues, we propose SAGML, an adaptive EVC framework via affective heterogeneous graph and multi-task language modeling. Instead of treating the emotion prior as a discrete tree, SAGML constructs a soft affective heterogeneous graph containing catalog-level emotion nodes and lexical-level emotion word nodes. The soft gate is injected into video-to-emotion graph attention as a continuous bias, allowing visually supported lexical emotions to remain recoverable rather than being removed by a hard mask. The resulting affective representation is fed together with visual tokens into a causal language decoder, while dual catalog and lexical heads impose explicit emotion distribution learning on the prompt hidden states. The overall model is trained with a joint objective that combines autoregressive caption generation and emotion distribution supervision. SAGML provides an error-resilient and multi-emotion-aware baseline for EVC.

OPERA: Aligning Open-Ended Reasoning via Objective Perplexity-based Reinforcement Learning

Reinforcement Learning (RL) has enabled LLMs to excel in objective reasoning tasks such as mathematics and code generation. However, applying RL to open-ended tasks, such as creative writing, remains challenging because LLM-as-a-judge reward models often exhibit stylistic biases and positional inconsistencies, leading to unstable supervision. To address this, we propose OPERA (Objective Perplexity-based Reflective Alignment), which replaces unreliable external judges with intrinsic rewards derived from perplexity dynamics. Specifically, we derive an intrinsic reward signal from perplexity dynamics, quantifying uncertainty reduction at critical reflective states. During the cold-start phase, we introduce a data synthesis method that leverages carefully designed guiding words to generate diverse reasoning traces, along with perplexity-prioritized rollouts that utilize internal log-probabilities to identify logically consistent reasoning branches. This pipeline yields a large-scale dataset comprising 20,000 high-quality reasoning trajectories. Empirical evaluations consistently demonstrate the scalability and efficacy of our approach in alignment for open-ended tasks. Implementing OPERA on Qwen3-8B establishes a new state-of-the-art among open-source models, achieving parity with or surpassing proprietary models like Gemini2.5 and MiniMax-M2.5 in some open-ended tasks. The code is available at https://github.com/pangpang-xuan/OPERA.

ScalingAttention: Discovering Intrinsic Sparse Attention Topology for Video Diffusion Transformers

While Diffusion Transformers (DiTs) have revolutionized high-fidelity video generation, their reliance on 3D full attention creates a quadratic computational bottleneck. Existing sparse methods face a dilemma: dynamic pruning suffers from prohibitive runtime overhead and memory fragmentation, while static heuristics fail to capture fine-grained dependencies. In this work, we propose ScalingAttention, a training-free framework grounded in a key inductive bias: while individual activations are input-dependent, the high-mass attention regions for each head rapidly converge to a stable, prompt-agnostic Intrinsic Sparse Topology. This topology is weight-encoded, scale-invariant, and efficient to extract. ScalingAttention decouples topology discovery from sparsity control via: (1) WEST (Weight-Encoded Sparse Topology), which extracts a robust block-sparse prior mask offline to eliminate runtime search; (2) FAST (Fidelity-Aware Sensitivity Tuning), which adaptively tunes head-wise sparsity based on diffusion fidelity requirements. To ensure practical acceleration, we co-design a hardware-aligned bit-wise block-sparse kernel. Experiments on Wan2.1 show up to 1.90X end-to-end speedup with superior fidelity, establishing a new Pareto frontier over state-of-the-art baselines.

MooD: An Efficient VA-Driven Affective Image Editing Framework via Fine-Grained Semantic Control

Affective image editing (AIE) aims to edit visual content to evoke target emotions. However, existing methods often overlook inference efficiency and predominantly depend on discrete emotion representations, which to some extent limits their practical applicability and makes it challenging to capture complex and subtle human emotions. To tackle these gaps, we propose MooD, the first framework that directly leverages continuous Valence-Arousal (VA) values for fine-grained and efficient AIE. Specifically, we first introduce a VA-Aware retrieval strategy to bridge vague affective values and concrete visual semantics. Building upon this, MooD integrates visual transfer and semantic guidance to achieve controllable AIE. Furthermore, we construct AffectSet, a VA-annotated dataset to support model optimization and evaluation. Extensive qualitative and quantitative experimental results demonstrate that our MooD achieves superior performance in both affective controllability and visual fidelity while maintaining high efficiency. A series of ablation studies further reveal the crucial factors of our design. Our code and data will be made publicly open soon.

AeroRAG: Structured Multimodal Retrieval-Augmented LLM for Fine-Grained Aerial Visual Reasoning

Despite recent progress in multimodal large language models (MLLMs), reliable visual question answering in aerial scenes remains challenging. In such scenes, task-critical evidence is often carried by small objects, explicit quantities, coarse locations, and inter-object relations, whereas conventional dense visual-token representations are not well aligned with these structured semantics. To address this interface mismatch, we propose AeroRAG, a scene-graph-guided multimodal retrieval-augmented generation framework for visual question answering. The framework first converts an input image into structured visual knowledge, including object categories, quantities, spatial locations, and semantic relations, and then retrieves query-relevant semantic chunks to construct compact prompts for a text-based large language model. Rather than relying on direct reasoning over dense visual tokens, our method introduces a more explicit intermediate interface between perception and language reasoning. Experiments on the AUG aerial dataset and the general-domain VG-150 benchmark show consistent improvements over six strong MLLM baselines, with the largest gains observed in dense aerial scenes and relation-sensitive reasoning. We further evaluate the framework on VQAv2 to verify that the proposed interface remains compatible with standard visual reasoning settings. These results suggest that structured retrieval is a practical design direction for deployment-oriented and grounded visual reasoning systems.

AIM-Bench: Benchmarking and Improving Affective Image Manipulation via Fine-Grained Hierarchical Control

Affective Image Manipulation (AIM) aims to evoke specific emotions through targeted editing. Current image editing benchmarks primarily focus on object-level modifications in general scenarios, lacking the fine-grained granularity to capture affective dimensions. To bridge this gap, we introduce the first benchmark designed for AIM termed AIM-Bench. This benchmark is built upon a dual-path affective modeling scheme that integrates the Mikels emotion taxonomy with the Valence-Arousal-Dominance framework, enabling high-level semantic and fine-grained continuous manipulation. Through a hierarchical human-in-the-loop workflow, we finally curate 800 high-quality samples covering 8 emotional categories and 5 editing types. To effectively assess performance, we also design a composite evaluation suite combining rule-based and model-based metrics to holistically assess instruction consistency, aesthetics, and emotional expressiveness. Extensive evaluations reveal that current editing models face significant challenges, most notably a prevalent positivity bias, which stemming from inherent imbalances in training data distribution. To tackle this, we propose a scalable data engine utilizing an inverse repainting strategy to construct AIM-40k, a balanced instruction-tuning dataset comprising 40k samples. Concretely, we enhance raw affective images via generative redrawing to establish high-fidelity ground truths, and synthesize input images with divergent emotions and paired precise instructions. Fine-tuning a baseline model on AIM-40k yields a 9.15% relative improvement in overall performance, demonstrating the effectiveness of our AIM-40k. Our data and related code will be made open soon.

DiffVC: A Non-autoregressive Framework Based on Diffusion Model for Video Captioning

Current video captioning methods usually use an encoder-decoder structure to generate text autoregressively. However, autoregressive methods have inherent limitations such as slow generation speed and large cumulative error. Furthermore, the few non-autoregressive counterparts suffer from deficiencies in generation quality due to the lack of sufficient multimodal interaction modeling. Therefore, we propose a non-autoregressive framework based on Diffusion model for Video Captioning (DiffVC) to address these issues. Its parallel decoding can effectively solve the problems of generation speed and cumulative error. At the same time, our proposed discriminative conditional Diffusion Model can generate higher-quality textual descriptions. Specifically, we first encode the video into a visual representation. During training, Gaussian noise is added to the textual representation of the ground-truth caption. Then, a new textual representation is generated via the discriminative denoiser with the visual representation as a conditional constraint. Finally, we input the new textual representation into a non-autoregressive language model to generate captions. During inference, we directly sample noise from the Gaussian distribution for generation. Experiments on MSVD, MSR-VTT, and VATEX show that our method can outperform previous non-autoregressive methods and achieve comparable performance to autoregressive methods, e.g., it achieved a maximum improvement of 9.9 on the CIDEr and improvement of 2.6 on the B@4, while having faster generation speed. The source code will be available soon.

EPIR: An Efficient Patch Tokenization, Integration and Representation Framework for Micro-expression Recognition

Micro-expression recognition can obtain the real emotion of the individual at the current moment. Although deep learning-based methods, especially Transformer-based methods, have achieved impressive results, these methods have high computational complexity due to the large number of tokens in the multi-head self-attention. In addition, the existing micro-expression datasets are small-scale, which makes it difficult for Transformer-based models to learn effective micro-expression representations. Therefore, we propose a novel Efficient Patch tokenization, Integration and Representation framework (EPIR), which can balance high recognition performance and low computational complexity. Specifically, we first propose a dual norm shifted tokenization (DNSPT) module to learn the spatial relationship between neighboring pixels in the face region, which is implemented by a refined spatial transformation and dual norm projection. Then, we propose a token integration module to integrate partial tokens among multiple cascaded Transformer blocks, thereby reducing the number of tokens without information loss. Furthermore, we design a discriminative token extractor, which first improves the attention in the Transformer block to reduce the unnecessary focus of the attention calculation on self-tokens, and uses the dynamic token selection module (DTSM) to select key tokens, thereby capturing more discriminative micro-expression representations. We conduct extensive experiments on four popular public datasets (i.e., CASME II, SAMM, SMIC, and CAS(ME)3. The experimental results show that our method achieves significant performance gains over the state-of-the-art methods, such as 9.6% improvement on the CAS(ME)$^3$ dataset in terms of UF1 and 4.58% improvement on the SMIC dataset in terms of UAR metric.

TR-ICRL: Test-Time Rethinking for In-Context Reinforcement Learning

In-Context Reinforcement Learning (ICRL) enables Large Language Models (LLMs) to learn online from external rewards directly within the context window. However, a central challenge in ICRL is reward estimation, as models typically lack access to ground-truths during inference. To address this limitation, we propose Test-Time Rethinking for In-Context Reinforcement Learning (TR-ICRL), a novel ICRL framework designed for both reasoning and knowledge-intensive tasks. TR-ICRL operates by first retrieving the most relevant instances from an unlabeled evaluation set for a given query. During each ICRL iteration, LLM generates a set of candidate answers for every retrieved instance. Next, a pseudo-label is derived from this set through majority voting. This label then serves as a proxy to give reward messages and generate formative feedbacks, guiding LLM through iterative refinement. In the end, this synthesized contextual information is integrated with the original query to form a comprehensive prompt, with the answer determining through a final round of majority voting. TR-ICRL is evaluated on mainstream reasoning and knowledge-intensive tasks, where it demonstrates significant performance gains. Remarkably, TR-ICRL improves Qwen2.5-7B by 21.23% on average on MedQA and even 137.59% on AIME2024. Extensive ablation studies and analyses further validate the effectiveness and robustness of our approach. Our code is available at https://github.com/pangpang-xuan/TR_ICRL.