LycheeDecode: Accelerating Long-Context LLM Inference via Hybrid-Head Sparse Decoding

The proliferation of long-context large language models (LLMs) exposes a key bottleneck: the rapidly expanding key-value cache during decoding, which imposes heavy memory and latency costs. While recent approaches attempt to alleviate this by sharing a single set of crucial tokens across layers, such coarse-grained sharing undermines model performance by neglecting the functional diversity of attention heads. To address this, we propose LycheeDecode, an efficient decoding method centered on a fine-grained hybrid-head attention mechanism that employs a hardware-efficient top-k selection strategy. Specifically, the novel HardKuma-based mechanism partitions attention heads into a small subset of retrieval heads that dynamically identify crucial tokens and a majority of sparse heads that reuse them for efficient computation. Through extensive experiments on leading models like Llama3 and Qwen3 across diverse benchmarks for long-context understanding (e.g., LongBench, RULER) and complex reasoning (e.g., AIME24, OlympiadBench), we demonstrate that LycheeDecode achieves generative quality comparable to, and at times surpassing even the full-attention baseline. Crucially, this is accomplished with up to a 2.7x speedup at a 128K context length. By preserving the functional diversity of attention heads, our fine-grained strategy overcomes the performance bottlenecks of existing methods, providing a powerful and validated pathway to both efficient and high-quality long-context LLM inference.

Structured Episodic Event Memory

Current approaches to memory in Large Language Models (LLMs) predominantly rely on static Retrieval-Augmented Generation (RAG), which often results in scattered retrieval and fails to capture the structural dependencies required for complex reasoning. For autonomous agents, these passive and flat architectures lack the cognitive organization necessary to model the dynamic and associative nature of long-term interaction. To address this, we propose Structured Episodic Event Memory (SEEM), a hierarchical framework that synergizes a graph memory layer for relational facts with a dynamic episodic memory layer for narrative progression. Grounded in cognitive frame theory, SEEM transforms interaction streams into structured Episodic Event Frames (EEFs) anchored by precise provenance pointers. Furthermore, we introduce an agentic associative fusion and Reverse Provenance Expansion (RPE) mechanism to reconstruct coherent narrative contexts from fragmented evidence. Experimental results on the LoCoMo and LongMemEval benchmarks demonstrate that SEEM significantly outperforms baselines, enabling agents to maintain superior narrative coherence and logical consistency.

PhysBrain: Human Egocentric Data as a Bridge from Vision Language Models to Physical Intelligence

Robotic generalization relies on physical intelligence: the ability to reason about state changes, contact-rich interactions, and long-horizon planning under egocentric perception and action. However, most VLMs are trained primarily on third-person data, creating a fundamental viewpoint mismatch for humanoid robots. Scaling robot egocentric data collection remains impractical due to high cost and limited diversity, whereas large-scale human egocentric videos offer a scalable alternative that naturally capture rich interaction context and causal structure. The key challenge is to convert raw egocentric videos into structured and reliable embodiment training supervision. Accordingly, we propose an Egocentric2Embodiment translation pipeline that transforms first-person videos into multi-level, schema-driven VQA supervision with enforced evidence grounding and temporal consistency, enabling the construction of the Egocentric2Embodiment dataset (E2E-3M) at scale. An egocentric-aware embodied brain, termed PhysBrain, is obtained by training on the E2E-3M dataset. PhysBrain exhibits substantially improved egocentric understanding, particularly for planning on EgoThink. It provides an egocentric-aware initialization that enables more sample-efficient VLA fine-tuning and higher SimplerEnv success rates (53.9\%), demonstrating effective transfer from human egocentric supervision to downstream robot control.

SealD-NeRF: Interactive Pixel-Level Editing for Dynamic Scenes by Neural Radiance Fields

The widespread adoption of implicit neural representations, especially Neural Radiance Fields (NeRF), highlights a growing need for editing capabilities in implicit 3D models, essential for tasks like scene post-processing and 3D content creation. Despite previous efforts in NeRF editing, challenges remain due to limitations in editing flexibility and quality. The key issue is developing a neural representation that supports local edits for real-time updates. Current NeRF editing methods, offering pixel-level adjustments or detailed geometry and color modifications, are mostly limited to static scenes. This paper introduces SealD-NeRF, an extension of Seal-3D for pixel-level editing in dynamic settings, specifically targeting the D-NeRF network. It allows for consistent edits across sequences by mapping editing actions to a specific timeframe, freezing the deformation network responsible for dynamic scene representation, and using a teacher-student approach to integrate changes.

Psy-LLM: Scaling up Global Mental Health Psychological Services with AI-based Large Language Models

The demand for psychological counseling has grown significantly in recent years, particularly with the global outbreak of COVID-19, which has heightened the need for timely and professional mental health support. Online psychological counseling has emerged as the predominant mode of providing services in response to this demand. In this study, we propose the Psy-LLM framework, an AI-based system leveraging Large Language Models (LLMs) for question-answering in online psychological consultation. Our framework combines pre-trained LLMs with real-world professional Q&A from psychologists and extensively crawled psychological articles. The Psy-LLM framework serves as a front-end tool for healthcare professionals, allowing them to provide immediate responses and mindfulness activities to alleviate patient stress. Additionally, it functions as a screening tool to identify urgent cases requiring further assistance. We evaluated the framework using intrinsic metrics, such as perplexity, and extrinsic evaluation metrics, with human participant assessments of response helpfulness, fluency, relevance, and logic. The results demonstrate the effectiveness of the Psy-LLM framework in generating coherent and relevant answers to psychological questions. This article concludes by discussing the potential of large language models to enhance mental health support through AI technologies in online psychological consultation.

Fossil Image Identification using Deep Learning Ensembles of Data Augmented Multiviews

Identification of fossil species is crucial to evolutionary studies. Recent advances from deep learning have shown promising prospects in fossil image identification. However, the quantity and quality of labeled fossil images are often limited due to fossil preservation, conditioned sampling, and expensive and inconsistent label annotation by domain experts, which pose great challenges to the training of deep learning based image classification models. To address these challenges, we follow the idea of the wisdom of crowds and propose a novel multiview ensemble framework, which collects multiple views of each fossil specimen image reflecting its different characteristics to train multiple base deep learning models and then makes final decisions via soft voting. We further develop OGS method that integrates original, gray, and skeleton views under this framework to demonstrate the effectiveness. Experimental results on the fusulinid fossil dataset over five deep learning based milestone models show that OGS using three base models consistently outperforms the baseline using a single base model, and the ablation study verifies the usefulness of each selected view. Besides, OGS obtains the superior or comparable performance compared to the method under well-known bagging framework. Moreover, as the available training data decreases, the proposed framework achieves more performance gains compared to the baseline. Furthermore, a consistency test with two human experts shows that OGS obtains the highest agreement with both the labels of dataset and the two experts. Notably, this methodology is designed for general fossil identification and it is expected to see applications on other fossil datasets. The results suggest the potential application when the quantity and quality of labeled data are particularly restricted, e.g., to identify rare fossil images.

Visibility-Aware Pixelwise View Selection for Multi-View Stereo Matching

The performance of PatchMatch-based multi-view stereo algorithms depends heavily on the source views selected for computing matching costs. Instead of modeling the visibility of different views, most existing approaches handle occlusions in an ad-hoc manner. To address this issue, we propose a novel visibility-guided pixelwise view selection scheme in this paper. It progressively refines the set of source views to be used for each pixel in the reference view based on visibility information provided by already validated solutions. In addition, the Artificial Multi-Bee Colony (AMBC) algorithm is employed to search for optimal solutions for different pixels in parallel. Inter-colony communication is performed both within the same image and among different images. Fitness rewards are added to validated and propagated solutions, effectively enforcing the smoothness of neighboring pixels and allowing better handling of textureless areas. Experimental results on the DTU dataset show our method achieves state-of-the-art performance among non-learning-based methods and retrieves more details in occluded and low-textured regions.