TMD-Bench: A Multi-Level Evaluation Paradigm for Music-Dance Co-Generation

Unified audio-visual generation is rapidly gaining industrial and creative relevance, enabling applications in virtual production and interactive media. However, when moving from general audio-video synthesis to music-dance co-generation, the task becomes substantially harder: musical rhythm, phrasing, and accents must drive choreographic motion at fine temporal resolution, and such rhythmic coupling is not captured by unimodal metrics or generic audiovisual consistency scores used in current evaluation practice. We introduce TMD-Bench, a benchmark for text-driven music-dance co-generation that assesses systems across unimodal generation quality, instruction adherence, and cross-modal rhythmic alignment. The benchmark integrates computable physical metrics with perceptual multimodal judgments, and is supported by a curated rhythm-aligned music-dance dataset and a fine-grained Music Captioner for structured music semantics. TMD-Bench further reveals that (i) modern commercial audio-visual models, such as Veo 3 and Sora 2, produce high-quality music and video, while rhythmic coupling remains less consistently optimized and leaves room for improvement, and (ii) our unified baseline RhyJAM trained on rhythm-aligned data achieves competitive beat-level synchronization while maintaining competitive unimodal fidelity. This presents prospects for building next-generation music-dance models that explicitly optimize rhythmic and kinetic coherence.

A Progressive Training Strategy for Vision-Language Models to Counteract Spatio-Temporal Hallucinations in Embodied Reasoning

Vision-Language Models (VLMs) have made significant strides in static image understanding but continue to face critical hurdles in spatiotemporal reasoning. A major bottleneck is "multi-image reasoning hallucination", where a massive performance drop between forward and reverse temporal queries reveals a dependence on superficial shortcuts instead of genuine causal understanding. To mitigate this, we first develop a new Chain-of-Thought (CoT) dataset that decomposes intricate reasoning into detailed spatiotemporal steps and definitive judgments. Building on this, we present a progressive training framework: it initiates with supervised pre-training on our CoT dataset to instill logical structures, followed by fine-tuning with scalable weakly-labeled data for broader generalization. Our experiments demonstrate that this approach not only improves backbone accuracy but also slashes the forward-backward performance gap from over 70\% to only 6.53\%. This confirms the method's ability to develop authentic dynamic reasoning and reduce the inherent temporal biases of current VLMs.

From Perception to Planning: Evolving Ego-Centric Task-Oriented Spatiotemporal Reasoning via Curriculum Learning

Modern vision-language models achieve strong performance in static perception, but remain limited in the complex spatiotemporal reasoning required for embodied, egocentric tasks. A major source of failure is their reliance on temporal priors learned from passive video data, which often leads to spatiotemporal hallucinations and poor generalization in dynamic environments. To address this, we present EgoTSR, a curriculum-based framework for learning task-oriented spatiotemporal reasoning. EgoTSR is built on the premise that embodied reasoning should evolve from explicit spatial understanding to internalized task-state assessment and finally to long-horizon planning. To support this paradigm, we construct EgoTSR-Data, a large-scale dataset comprising 46 million samples organized into three stages: Chain-of-Thought (CoT) supervision, weakly supervised tagging, and long-horizon sequences. Extensive experiments demonstrate that EgoTSR effectively eliminates chronological biases, achieving 92.4% accuracy on long-horizon logical reasoning tasks while maintaining high fine-grained perceptual precision, significantly outperforming existing open-source and closed-source state-of-the-art models.

ImVideoEdit: Image-learning Video Editing via 2D Spatial Difference Attention Blocks

Current video editing models often rely on expensive paired video data, which limits their practical scalability. In essence, most video editing tasks can be formulated as a decoupled spatiotemporal process, where the temporal dynamics of the pretrained model are preserved while spatial content is selectively and precisely modified. Based on this insight, we propose ImVideoEdit, an efficient framework that learns video editing capabilities entirely from image pairs. By freezing the pre-trained 3D attention modules and treating images as single-frame videos, we decouple the 2D spatial learning process to help preserve the original temporal dynamics. The core of our approach is a Predict-Update Spatial Difference Attention module that progressively extracts and injects spatial differences. Rather than relying on rigid external masks, we incorporate a Text-Guided Dynamic Semantic Gating mechanism for adaptive and implicit text-driven modifications. Despite training on only 13K image pairs for 5 epochs with exceptionally low computational overhead, ImVideoEdit achieves editing fidelity and temporal consistency comparable to larger models trained on extensive video datasets.

SpatialReward: Verifiable Spatial Reward Modeling for Fine-Grained Spatial Consistency in Text-to-Image Generation

Recent advances in text-to-image (T2I) generation via reinforcement learning (RL) have benefited from reward models that assess semantic alignment and visual quality. However, most existing reward models pay limited attention to fine-grained spatial relationships, often producing images that appear plausible overall yet contain inaccuracies in object positioning. In this work, we present \textbf{SpatialReward}, a verifiable reward model explicitly designed to evaluate spatial layouts in generated images. SpatialReward adopts a multi-stage pipeline: a \emph{Prompt Decomposer} extracts entities, attributes, and spatial metadata from free-form prompts; expert detectors provide accurate visual grounding of object positions and attributes; and a vision-language model applies chain-of-thought reasoning over grounded observations to assess complex spatial relations that are challenging for rule-based methods. To more comprehensively evaluate spatial relationships in generated images, we introduce \textbf{SpatRelBench}, a benchmark covering object attributes, orientation, inter-object relations, and rendered text placement. Experiments on Stable Diffusion and FLUX show that incorporating SpatialReward into RL training consistently improves spatial consistency and overall generation quality, with results aligned more closely to human judgments. These findings indicate that verifiable reward models hold considerable potential for enabling more accurate and controllable optimization in text-to-image generation models.

CronusVLA: Transferring Latent Motion Across Time for Multi-Frame Prediction in Manipulation

Recent vision-language-action (VLA) models built on pretrained vision-language models (VLMs) have demonstrated strong generalization across manipulation tasks. However, they remain constrained by a single-frame observation paradigm and cannot fully benefit from the motion information offered by aggregated multi-frame historical observations, as the large vision-language backbone introduces substantial computational cost and inference latency. We propose CronusVLA, a unified framework that extends single-frame VLA models to the multi-frame paradigm through an efficient post-training stage. CronusVLA comprises three key components: (1) single-frame pretraining on large-scale embodied datasets with autoregressive action tokens prediction, which establishes an embodied vision-language foundation; (2) multi-frame encoding, adapting the prediction of vision-language backbones from discrete action tokens to motion features during post-training, and aggregating motion features from historical frames into a feature chunking; (3) cross-frame decoding, which maps the feature chunking to accurate actions via a shared decoder with cross-attention. By reducing redundant token computation and caching past motion features, CronusVLA achieves efficient inference. As an application of motion features, we further propose an action adaptation mechanism based on feature-action retrieval to improve model performance during finetuning. CronusVLA achieves state-of-the-art performance on SimplerEnv with 70.9% success rate, and 12.7% improvement over OpenVLA on LIBERO. Real-world Franka experiments also show the strong performance and robustness.

Vela: Scalable Embeddings with Voice Large Language Models for Multimodal Retrieval

Multimodal large language models (MLLMs) have seen substantial progress in recent years. However, their ability to represent multimodal information in the acoustic domain remains underexplored. In this work, we introduce Vela, a novel framework designed to adapt MLLMs for the generation of universal multimodal embeddings. By leveraging MLLMs with specially crafted prompts and selected in-context learning examples, Vela effectively bridges the modality gap across various modalities. We then propose a single-modality training approach, where the model is trained exclusively on text pairs. Our experiments show that Vela outperforms traditional CLAP models in standard text-audio retrieval tasks. Furthermore, we introduce new benchmarks that expose CLAP models' limitations in handling long texts and complex retrieval tasks. In contrast, Vela, by harnessing the capabilities of MLLMs, demonstrates robust performance in these scenarios. Our code will soon be available.

Speech Token Prediction via Compressed-to-fine Language Modeling for Speech Generation

Neural audio codecs, used as speech tokenizers, have demonstrated remarkable potential in the field of speech generation. However, to ensure high-fidelity audio reconstruction, neural audio codecs typically encode audio into long sequences of speech tokens, posing a significant challenge for downstream language models in long-context modeling. We observe that speech token sequences exhibit short-range dependency: due to the monotonic alignment between text and speech in text-to-speech (TTS) tasks, the prediction of the current token primarily relies on its local context, while long-range tokens contribute less to the current token prediction and often contain redundant information. Inspired by this observation, we propose a \textbf{compressed-to-fine language modeling} approach to address the challenge of long sequence speech tokens within neural codec language models: (1) \textbf{Fine-grained Initial and Short-range Information}: Our approach retains the prompt and local tokens during prediction to ensure text alignment and the integrity of paralinguistic information; (2) \textbf{Compressed Long-range Context}: Our approach compresses long-range token spans into compact representations to reduce redundant information while preserving essential semantics. Extensive experiments on various neural audio codecs and downstream language models validate the effectiveness and generalizability of the proposed approach, highlighting the importance of token compression in improving speech generation within neural codec language models. The demo of audio samples will be available at https://anonymous.4open.science/r/SpeechTokenPredictionViaCompressedToFinedLM.

Diff-Prompt: Diffusion-Driven Prompt Generator with Mask Supervision

Prompt learning has demonstrated promising results in fine-tuning pre-trained multimodal models. However, the performance improvement is limited when applied to more complex and fine-grained tasks. The reason is that most existing methods directly optimize the parameters involved in the prompt generation process through loss backpropagation, which constrains the richness and specificity of the prompt representations. In this paper, we propose Diffusion-Driven Prompt Generator (Diff-Prompt), aiming to use the diffusion model to generate rich and fine-grained prompt information for complex downstream tasks. Specifically, our approach consists of three stages. In the first stage, we train a Mask-VAE to compress the masks into latent space. In the second stage, we leverage an improved Diffusion Transformer (DiT) to train a prompt generator in the latent space, using the masks for supervision. In the third stage, we align the denoising process of the prompt generator with the pre-trained model in the semantic space, and use the generated prompts to fine-tune the model. We conduct experiments on a complex pixel-level downstream task, referring expression comprehension, and compare our method with various parameter-efficient fine-tuning approaches. Diff-Prompt achieves a maximum improvement of 8.87 in R@1 and 14.05 in R@5 compared to the foundation model and also outperforms other state-of-the-art methods across multiple metrics. The experimental results validate the effectiveness of our approach and highlight the potential of using generative models for prompt generation. Code is available at https://github.com/Kelvin-ywc/diff-prompt.

EyecareGPT: Boosting Comprehensive Ophthalmology Understanding with Tailored Dataset, Benchmark and Model

Medical Large Vision-Language Models (Med-LVLMs) demonstrate significant potential in healthcare, but their reliance on general medical data and coarse-grained global visual understanding limits them in intelligent ophthalmic diagnosis. Currently, intelligent ophthalmic diagnosis faces three major challenges: (i) Data. The lack of deeply annotated, high-quality, multi-modal ophthalmic visual instruction data; (ii) Benchmark. The absence of a comprehensive and systematic benchmark for evaluating diagnostic performance; (iii) Model. The difficulty of adapting holistic visual architectures to fine-grained, region-specific ophthalmic lesion identification. In this paper, we propose the Eyecare Kit, which systematically tackles the aforementioned three key challenges with the tailored dataset, benchmark and model: First, we construct a multi-agent data engine with real-life ophthalmology data to produce Eyecare-100K, a high-quality ophthalmic visual instruction dataset. Subsequently, we design Eyecare-Bench, a benchmark that comprehensively evaluates the overall performance of LVLMs on intelligent ophthalmic diagnosis tasks across multiple dimensions. Finally, we develop the EyecareGPT, optimized for fine-grained ophthalmic visual understanding thoroughly, which incorporates an adaptive resolution mechanism and a layer-wise dense connector. Extensive experimental results indicate that the EyecareGPT achieves state-of-the-art performance in a range of ophthalmic tasks, underscoring its significant potential for the advancement of open research in intelligent ophthalmic diagnosis. Our project is available at https://github.com/DCDmllm/EyecareGPT.