arXiv subjects

Feng Ling

Publications and source records attributed to Feng Ling.

15 recordsLinked to original sources

arXiv2026

ToST: A Tree-of-Thought Socratic Teaching Framework for Multi-Path Guidance and Parallel Thinking

Large Language Models (LLMs) exhibit strong problem-solving abilities, positioning them as promising agents for Socratic teaching to guide students through step-by-step heuristic questioning. However, existing approaches typically adopt a one-problem-one-solution paradigm, restricting the teaching guidance to a single linear reasoning path. This design limits instructional flexibility, weakens error recovery, and restricts students' ability to engage in parallel thinking to explore multiple valid solutions. To overcome these, we propose ToST, a Tree-of-Thought Socratic Teaching framework that explicitly supports multi-path guidance under a one-problem-multiple-solutions paradigm. ToST employs Parallel Sowing, a parallel-thinking-oriented questioning strategy to encourage students to approach problems from diverse perspectives, and a Multi-Path Adaptive Guidance mechanism to provide more robust and non-linear instructions across alternative solution trajectories. Concurrently, to fill the void in systematically evaluating such non-linear instructional capabilities, we advance the task of multi-path Socratic guidance by establishing MPSG-Bench, a comprehensive benchmark that includes a dataset of 31K multi-path teaching dialogues and a five-dimensional evaluation framework grounded in the SOLO (Structure of Observed Learning Outcomes) theory to assess parallel-thinking guidance. Experimental results demonstrate that ToST significantly enhances guidance success rates while empowering students to navigate and explore multiple solution paths more effectively under both automatic and human metrics.

cs.AI↗

arXiv2026

Cilia-driven transport in confined ducts: an active porous media model

Ciliated organs transport viscous fluids through confined ducts, yet how duct morphology and ciliary activity jointly set the limits of flow rate and sustainable pressure remains unclear. Here, we model dense arrays of beating cilia lining duct walls as an active porous medium driven by prescribed metachronal waves, and identify two key morphological parameters that govern transport: the ciliary confinement ratio and the mean ciliary fraction. The resulting flows are described by the incompressible Navier-Stokes-Brinkman equations, which we solve numerically using a spectral method in the low-Reynolds-number regime. We also develop a complementary mean-field analytical model. The active porous medium framework provides an intermediate description between classical envelope theories and filament-resolved simulations and enables a systematic investigation of how fluid transport is shaped by confinement and packing of ciliary material. We find that transport is characterized by a decreasing linear relationship between flow rate and pressure generation, marking a fundamental trade-off between throughput and sustainable adverse pressure. These results provide a unified physical interpretation of the morphological diversity of ciliated ducts, from high-throughput ciliary carpets to pressure-generating ciliary flames, and offer guiding principles for the design of bio-inspired microfluidic pumps.

physics.flu-dyn↗

arXiv2026

Seedance 2.0: Advancing Video Generation for World Complexity

Seedance 2.0 is a new native multi-modal audio-video generation model, officially released in China in early February 2026. Compared with its predecessors, Seedance 1.0 and 1.5 Pro, Seedance 2.0 adopts a unified, highly efficient, and large-scale architecture for multi-modal audio-video joint generation. This allows it to support four input modalities: text, image, audio, and video, by integrating one of the most comprehensive suites of multi-modal content reference and editing capabilities available in the industry to date. It delivers substantial, well-rounded improvements across all key sub-dimensions of video and audio generation. In both expert evaluations and public user tests, the model has demonstrated performance on par with the leading levels in the field. Seedance 2.0 supports direct generation of audio-video content with durations ranging from 4 to 15 seconds, with native output resolutions of 480p and 720p. For multi-modal inputs as reference, its current open platform supports up to 3 video clips, 9 images, and 3 audio clips. In addition, we provide Seedance 2.0 Fast version, an accelerated variant of Seedance 2.0 designed to boost generation speed for low-latency scenarios. Seedance 2.0 has delivered significant improvements to its foundational generation capabilities and multi-modal generation performance, bringing an enhanced creative experience for end users.

cs.CV↗

arXiv2026

Flow caching for autoregressive video generation

Autoregressive models, often built on Transformer architectures, represent a powerful paradigm for generating ultra-long videos by synthesizing content in sequential chunks. However, this sequential generation process is notoriously slow. While caching strategies have proven effective for accelerating traditional video diffusion models, existing methods assume uniform denoising across all frames-an assumption that breaks down in autoregressive models where different video chunks exhibit varying similarity patterns at identical timesteps. In this paper, we present FlowCache, the first caching framework specifically designed for autoregressive video generation. Our key insight is that each video chunk should maintain independent caching policies, allowing fine-grained control over which chunks require recomputation at each timestep. We introduce a chunkwise caching strategy that dynamically adapts to the unique denoising characteristics of each chunk, complemented by a joint importance-redundancy optimized KV cache compression mechanism that maintains fixed memory bounds while preserving generation quality. Our method achieves remarkable speedups of 2.38 times on MAGI-1 and 6.7 times on SkyReels-V2, with negligible quality degradation (VBench: 0.87 increase and 0.79 decrease respectively). These results demonstrate that FlowCache successfully unlocks the potential of autoregressive models for real-time, ultra-long video generation-establishing a new benchmark for efficient video synthesis at scale. The code is available at https://github.com/mikeallen39/FlowCache.

cs.CV↗

arXiv2025

Seedance 1.5 pro: A Native Audio-Visual Joint Generation Foundation Model

Recent strides in video generation have paved the way for unified audio-visual generation. In this work, we present Seedance 1.5 pro, a foundational model engineered specifically for native, joint audio-video generation. Leveraging a dual-branch Diffusion Transformer architecture, the model integrates a cross-modal joint module with a specialized multi-stage data pipeline, achieving exceptional audio-visual synchronization and superior generation quality. To ensure practical utility, we implement meticulous post-training optimizations, including Supervised Fine-Tuning (SFT) on high-quality datasets and Reinforcement Learning from Human Feedback (RLHF) with multi-dimensional reward models. Furthermore, we introduce an acceleration framework that boosts inference speed by over 10X. Seedance 1.5 pro distinguishes itself through precise multilingual and dialect lip-syncing, dynamic cinematic camera control, and enhanced narrative coherence, positioning it as a robust engine for professional-grade content creation. Seedance 1.5 pro is now accessible on Volcano Engine at https://console.volcengine.com/ark/region:ark+cn-beijing/experience/vision?type=GenVideo.

cs.CV↗

arXiv2025

PAROAttention: Pattern-Aware ReOrdering for Efficient Sparse and Quantized Attention in Visual Generation Models

In visual generation, the quadratic complexity of attention mechanisms results in high memory and computational costs, especially for longer token sequences required in high-resolution image or multi-frame video generation. To address this, prior research has explored techniques such as sparsification and quantization. However, these techniques face significant challenges under low density and reduced bitwidths. Through systematic analysis, we identify that the core difficulty stems from the dispersed and irregular characteristics of visual attention patterns. Therefore, instead of introducing specialized sparsification and quantization design to accommodate such patterns, we propose an alternative strategy: *reorganizing* the attention pattern to alleviate the challenges. Inspired by the local aggregation nature of visual feature extraction, we design a novel **Pattern-Aware token ReOrdering (PARO)** technique, which unifies the diverse attention patterns into a hardware-friendly block-wise pattern. This unification substantially simplifies and enhances both sparsification and quantization. We evaluate the performance-efficiency trade-offs of various design choices and finalize a methodology tailored for the unified pattern. Our approach, **PAROAttention**, achieves video and image generation with lossless metrics, and nearly identical results from full-precision (FP) baselines, while operating at notably lower density (~20%-30%) and bitwidth (**INT8/INT4**), achieving a **1.9x** to **2.7x** end-to-end latency speedup.

cs.CV↗

arXiv2025

Seaweed-7B: Cost-Effective Training of Video Generation Foundation Model

This technical report presents a cost-efficient strategy for training a video generation foundation model. We present a mid-sized research model with approximately 7 billion parameters (7B) called Seaweed-7B trained from scratch using 665,000 H100 GPU hours. Despite being trained with moderate computational resources, Seaweed-7B demonstrates highly competitive performance compared to contemporary video generation models of much larger size. Design choices are especially crucial in a resource-constrained setting. This technical report highlights the key design decisions that enhance the performance of the medium-sized diffusion model. Empirically, we make two observations: (1) Seaweed-7B achieves performance comparable to, or even surpasses, larger models trained on substantially greater GPU resources, and (2) our model, which exhibits strong generalization ability, can be effectively adapted across a wide range of downstream applications either by lightweight fine-tuning or continue training. See the project page at https://seaweed.video/

cs.CV↗

arXiv2024

Deep Dive into Model-free Reinforcement Learning for Biological and Robotic Systems: Theory and Practice

Animals and robots exist in a physical world and must coordinate their bodies to achieve behavioral objectives. With recent developments in deep reinforcement learning, it is now possible for scientists and engineers to obtain sensorimotor strategies (policies) for specific tasks using physically simulated bodies and environments. However, the utility of these methods goes beyond the constraints of a specific task; they offer an exciting framework for understanding the organization of an animal sensorimotor system in connection to its morphology and physical interaction with the environment, as well as for deriving general design rules for sensing and actuation in robotic systems. Algorithms and code implementing both learning agents and environments are increasingly available, but the basic assumptions and choices that go into the formulation of an embodied feedback control problem using deep reinforcement learning may not be immediately apparent. Here, we present a concise exposition of the mathematical and algorithmic aspects of model-free reinforcement learning, specifically through the use of \textit{actor-critic} methods, as a tool for investigating the feedback control underlying animal and robotic behavior.

cs.RO↗

arXiv2024

AffineQuant: Affine Transformation Quantization for Large Language Models

The significant resource requirements associated with Large-scale Language Models (LLMs) have generated considerable interest in the development of techniques aimed at compressing and accelerating neural networks. Among these techniques, Post-Training Quantization (PTQ) has emerged as a subject of considerable interest due to its noteworthy compression efficiency and cost-effectiveness in the context of training. Existing PTQ methods for LLMs limit the optimization scope to scaling transformations between pre- and post-quantization weights. In this paper, we advocate for the direct optimization using equivalent Affine transformations in PTQ (AffineQuant). This approach extends the optimization scope and thus significantly minimizing quantization errors. Additionally, by employing the corresponding inverse matrix, we can ensure equivalence between the pre- and post-quantization outputs of PTQ, thereby maintaining its efficiency and generalization capabilities. To ensure the invertibility of the transformation during optimization, we further introduce a gradual mask optimization method. This method initially focuses on optimizing the diagonal elements and gradually extends to the other elements. Such an approach aligns with the Levy-Desplanques theorem, theoretically ensuring invertibility of the transformation. As a result, significant performance improvements are evident across different LLMs on diverse datasets. To illustrate, we attain a C4 perplexity of 15.76 (2.26 lower vs 18.02 in OmniQuant) on the LLaMA2-7B model of W4A4 quantization without overhead. On zero-shot tasks, AffineQuant achieves an average of 58.61 accuracy (1.98 lower vs 56.63 in OmniQuant) when using 4/4-bit quantization for LLaMA-30B, which setting a new state-of-the-art benchmark for PTQ in LLMs.

cs.LG↗

arXiv2023

Utilizing entropy to systematically quantify the resting-condition baroreflex regulation function

Baroreflex is critical to maintain the blood pressure homeostasis, and the quantification of the baroreflex regulation function (BRF) can provide guidance for disease diagnosis, treatment and healthcare. Current quantification of the BRF such as baroreflex sensitivity cannot represent the BRF systematically. From the perspective of complex systems, we regard that the BRF is the emergence result of the diverse states and interactions in the physiological mechanisms. Therefore, the three-layer emergence is constructed in this work, which is from the physiological mechanisms to the physiological indexes and then to the BRF. On this basis, since the entropy in statistical physics macroscopically measures the diversity of the system's states, a new index called the PhysioEnt is proposed to represent the BRF and quantify the physical relationships between the BRF and four physiological indexes, baroreflex sensitivity, heart rate, heart rate variability, and systolic blood pressure. Based on the proposed method, some new findings with medical significance are obtained, including the mechanisms that aging and obesity affect the resting-condition BRF are different, and the resting-condition BRFs of men and older people depend more on the physiological processes among organs/tissues. Based on the measurable indexes, the proposed method would support the individualized medicine prospectively.

physics.med-ph↗

arXiv2022

Spontaneous phase coordination and fluid pumping in model ciliary carpets

Ciliated tissues such as in the mammalian lungs, brains, and reproductive tracts, are specialized to pump fluid. They generate flows by the collective activity of hundreds of thousands of individual cilia that beat in a striking metachronal wave pattern. Despite progress in analyzing cilia coordination, a general theory that links coordination and fluid pumping in the limit of large arrays of cilia remains lacking. Here, we conduct in-silico experiments with thousands of hydrodynamically-interacting cilia, and we develop a continuum theory in the limit of infinitely-many independently beating cilia by combining tools from active matter and classical Stokes flow. We find, in both simulations and theory, that isotropic and synchronized ciliary states are unstable. Traveling waves emerge regardless of initial conditions, but the characteristics of the wave and net flows depend on cilia and tissue properties. That is, metachronal phase coordination is a stable global attractor in large ciliary carpets, even under finite perturbations to cilia and tissue properties. These results support the notion that functional specificity of ciliated tissues is interlaced with the tissue architecture and cilia beat kinematics and open up the prospect of establishing structure-to-function maps from cilium-level beat to tissue-level coordination and fluid pumping.

cond-mat.soft↗

arXiv2021

A free lunch from ViT:Adaptive Attention Multi-scale Fusion Transformer for Fine-grained Visual Recognition

Learning subtle representation about object parts plays a vital role in fine-grained visual recognition (FGVR) field. The vision transformer (ViT) achieves promising results on computer vision due to its attention mechanism. Nonetheless, with the fixed size of patches in ViT, the class token in deep layer focuses on the global receptive field and cannot generate multi-granularity features for FGVR. To capture region attention without box annotations and compensate for ViT shortcomings in FGVR, we propose a novel method named Adaptive attention multi-scale Fusion Transformer (AFTrans). The Selective Attention Collection Module (SACM) in our approach leverages attention weights in ViT and filters them adaptively to correspond with the relative importance of input patches. The multiple scales (global and local) pipeline is supervised by our weights sharing encoder and can be easily trained end-to-end. Comprehensive experiments demonstrate that AFTrans can achieve SOTA performance on three published fine-grained benchmarks: CUB-200-2011, Stanford Dogs and iNat2017.

cs.CV↗

arXiv2020

Learning to swim in potential flow

Fish swim by undulating their bodies. These propulsive motions require coordinated shape changes of a body that interacts with its fluid environment, but the specific shape coordination that leads to robust turning and swimming motions remains unclear. To address the problem of underwater motion planning, we propose a simple model of a three-link fish swimming in a potential flow environment and we use model-free reinforcement learning for shape control. We arrive at optimal shape changes for two swimming tasks: swimming in a desired direction and swimming towards a known target. This fish model belongs to a class of problems in geometric mechanics, known as driftless dynamical systems, which allow us to analyze the swimming behavior in terms of geometric phases over the shape space of the fish. These geometric methods are less intuitive in the presence of drift. Here, we use the shape space analysis as a tool for assessing, visualizing, and interpreting the control policies obtained via reinforcement learning in the absence of drift. We then examine the robustness of these policies to drift-related perturbations. Although the fish has no direct control over the drift itself, it learns to take advantage of the presence of moderate drift to reach its target.

q-bio.QM↗

arXiv2020

Cloud detection in Landsat-8 imagery in Google Earth Engine based on a deep neural network

Google Earth Engine (GEE) provides a convenient platform for applications based on optical satellite imagery of large areas. With such data sets, the detection of cloud is often a necessary prerequisite step. Recently, deep learning-based cloud detection methods have shown their potential for cloud detection but they can only be applied locally, leading to inefficient data downloading time and storage problems. This letter proposes a method to directly perform cloud detection in Landsat-8 imagery in GEE based on deep learning (DeepGEE-CD). A deep neural network (DNN) was first trained locally, and then the trained DNN was deployed in the JavaScript client of GEE. An experiment was undertaken to validate the proposed method with a set of Landsat-8 images and the results show that DeepGEE-CD outperformed the widely used function of mask (Fmask) algorithm. The proposed DeepGEE-CD approach can accurately detect cloud in Landsat-8 imagery without downloading it, making it a promising method for routine cloud detection of Landsat-8 imagery in GEE.

eess.IV↗

arXiv2018

Instability-driven Oscillations of Elastic Microfilaments

Cilia and flagella are highly conserved slender organelles that exhibit a variety of rhythmic beating patterns from non-planar cone-like motions to planar wave-like deformations. Although their internal structure, composed of a microtubule-based axoneme driven by dynein motors, is known, the mechanism responsible for these beating patterns remains elusive. Existing theories suggest that the dynein activity is dynamically regulated, via a geometric feedback from the cilium's mechanical deformation to the dynein force. An alternative, open-loop mechanism based on a 'flutter' instability was recently proven to lead to planar oscillations of elastic filaments under follower forces. Here, we show that an elastic filament in viscous fluid, clamped at one end and acted on by an external distribution of compressive axial forces, exhibits a Hopf bifurcation that leads to non-planar spinning of the buckled filament at a locked curvature. We also show the existence of a second bifurcation, at larger force values, that induces a transition from non-planar spinning to planar wave-like oscillations. We elucidate the nature of these instabilities using a combination of nonlinear numerical analysis, linear stability theory, and low-order bead-spring models. Our results show that away from the transition thresholds, these beating patterns are robust to perturbations in the distribution of axial forces and in the filament configuration. These findings support the theory that an open-loop, instability-driven mechanism could explain both the sustained oscillations and the wide variety of periodic beating patterns observed in cilia and flagella.

physics.flu-dyn↗