arXiv subjects

Mingyu Yang

Publications and source records attributed to Mingyu Yang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Zebra-Llama: Towards Extremely Efficient Hybrid Models

With the growing demand for deploying large language models (LLMs) across diverse applications, improving their inference efficiency is crucial for sustainable and democratized access. However, retraining LLMs to meet new user-specific requirements is prohibitively expensive and environmentally unsustainable. In this work, we propose a practical and scalable alternative: composing efficient hybrid language models from existing pre-trained models. Our approach, Zebra-Llama, introduces a family of 1B, 3B, and 8B hybrid models by combining State Space Models (SSMs) and Multi-head Latent Attention (MLA) layers, using a refined initialization and post-training pipeline to efficiently transfer knowledge from pre-trained Transformers. Zebra-Llama achieves Transformer-level accuracy with near-SSM efficiency using only 7-11B training tokens (compared to trillions of tokens required for pre-training) and an 8B teacher. Moreover, Zebra-Llama dramatically reduces KV cache size -down to 3.9%, 2%, and 2.73% of the original for the 1B, 3B, and 8B variants, respectively-while preserving 100%, 100%, and >97% of average zero-shot performance on LM Harness tasks. Compared to models like MambaInLLaMA, X-EcoMLA, Minitron, and Llamba, Zebra-Llama consistently delivers competitive or superior accuracy while using significantly fewer tokens, smaller teachers, and vastly reduced KV cache memory. Notably, Zebra-Llama-8B surpasses Minitron-8B in few-shot accuracy by 7% while using 8x fewer training tokens, over 12x smaller KV cache, and a smaller teacher (8B vs. 15B). It also achieves 2.6x-3.8x higher throughput (tokens/s) than MambaInLlama up to a 32k context length. We will release code and model checkpoints upon acceptance.

cs.LG↗

arXiv2025

Elastic Index Selection for Label-Hybrid AKNN Search

Real-world vector embeddings are usually associated with extra labels, such as attributes and keywords. Many applications require the nearest neighbor search that contains specific labels, such as searching for product image embeddings restricted to a particular brand. A straightforward approach is to materialize all possible indices according to the complete query label workload. However, this leads to an exponential increase in both index space and processing time, which significantly limits scalability and efficiency. In this paper, we leverage the inclusion relationships among query label sets to construct partial indexes, enabling index sharing across queries for improved construction efficiency. We introduce \textit{elastic factor} bounds to guarantee search performance and use the greedy algorithm to select indices that meet the bounds, achieving a tradeoff between efficiency and space. Meanwhile, we also designed the algorithm to achieve the best elastic factor under a given space limitation. Experimental results on multiple real datasets demonstrate that our algorithm can achieve near-optimal search performance, achieving up to 10x-500x search efficiency speed up over state-of-the-art approaches. Our algorithm is highly versatile, since it is not constrained by index type and can seamlessly integrate with existing optimized libraries.

cs.DB↗

arXiv2025

Emergent Chiral Spin Crystal Phase in (111) SrRuO3 Thin Films

Perovskite ruthenates are fascinating playgrounds for exploring topological spin textures, but generally rely on extrinsic mechanisms to trigger the noncoplanar states. Here we report the discovery of an emergent chiral spin crystal phase in (111) SrRuO3 epitaxial films, characterized by a significant topological Hall effect and noncoplanar spin arrangements with different propagation vectors along two orthogonal directions. Instead of driven by the enhanced Dzyaloshinskii-Moriya interaction due to broken inversion symmetry at heterointerfaces, this emergent state arises intrinsically from the interplay of dipolar interactions and magnetic frustration, leading to the stabilization of topological phases in much thicker films. These findings open a new pathway for creating and controlling the topological spin states in perovskites, with broad implications for spintronic device design.

cond-mat.str-el↗

arXiv2025

DSDE: Dynamic Speculative Decoding with KLD Stability for Real-World Serving

Speculative decoding accelerates large language model inference, but its reliance on a fixed speculation length is suboptimal in large-batch serving environments with diverse requests. This paper explores a new direction for dynamic adaptation by investigating a novel class of post-hoc, diagnostic signals. We propose Dynamic Speculative Decoding Engine (DSDE), a training-free framework built on two primary components: (1) a predictive signal based on the variance of the Kullback-Leibler (KLD) divergence, which diagnoses the generation's regional stability, and (2) an adaptive speculation length cap to mitigate the straggler problem in per-sequence decoding. Experiments demonstrate the potential of using KLD-based stability signals for dynamic adaptation. An algorithm guided by these signals achieves end-to-end latency competitive with leading baselines and exhibits superior robustness across diverse workloads. This robustness is particularly valuable in challenging low-acceptance-rate regimes, where the proposed signal maintains its diagnostic utility. Collectively, these findings validate post-hoc signals as a valuable component for building more robust and intelligent LLM inference systems, and highlight a promising direction for future research on dynamic speculation length adaptation.

cs.DC↗

arXiv2025

X-EcoMLA: Upcycling Pre-Trained Attention into MLA for Efficient and Extreme KV Compression

Multi-head latent attention (MLA) is designed to optimize KV cache memory through low-rank key-value joint compression. Rather than caching keys and values separately, MLA stores their compressed latent representations, reducing memory overhead while maintaining the performance. While MLA improves memory efficiency without compromising language model accuracy, its major limitation lies in its integration during the pre-training phase, requiring models to be trained from scratch. This raises a key question: can we use MLA's benefits fully or partially in models that have already been pre-trained with different attention mechanisms? In this paper, we propose X-EcoMLA to deploy post training distillation to enable the upcycling of Transformer-based attention into an efficient hybrid MLA variant through lightweight post-training adaptation, bypassing the need for extensive pre-training. We demonstrate that leveraging the dark knowledge of a well-trained model can enhance training accuracy and enable extreme KV cache compression in MLA without compromising model performance. The experimental results show that our proposed method can effectively compress the KV cache while preserving the performance on the benchmarks; specifically, for Llama3.2-1B-Instruct baseline, a 6.4x compression achieves the same average score by using only 3.6B training tokens and 70 GPU hours on AMD MI300, whereas a 10.6x compression have less than 0.1% average score drop with 7B training tokens and 140 GPU hours. The code for this work is available at https://github.com/AMD-AGI/AMD-Hybrid-Models.

cs.CL↗

arXiv2025

Yan: Foundational Interactive Video Generation

We present Yan, a foundational framework for interactive video generation, covering the entire pipeline from simulation and generation to editing. Specifically, Yan comprises three core modules. AAA-level Simulation: We design a highly-compressed, low-latency 3D-VAE coupled with a KV-cache-based shift-window denoising inference process, achieving real-time 1080P/60FPS interactive simulation. Multi-Modal Generation: We introduce a hierarchical autoregressive caption method that injects game-specific knowledge into open-domain multi-modal video diffusion models (VDMs), then transforming the VDM into a frame-wise, action-controllable, real-time infinite interactive video generator. Notably, when the textual and visual prompts are sourced from different domains, the model demonstrates strong generalization, allowing it to blend and compose the style and mechanics across domains flexibly according to user prompts. Multi-Granularity Editing: We propose a hybrid model that explicitly disentangles interactive mechanics simulation from visual rendering, enabling multi-granularity video content editing during interaction through text. Collectively, Yan offers an integration of these modules, pushing interactive video generation beyond isolated capabilities toward a comprehensive AI-driven interactive creation paradigm, paving the way for the next generation of creative tools, media, and entertainment. The project page is: https://greatx3.github.io/Yan/.

cs.CV↗

arXiv2025

VSAG: An Optimized Search Framework for Graph-based Approximate Nearest Neighbor Search

Approximate nearest neighbor search (ANNS) is a fundamental problem in vector databases and AI infrastructures. Recent graph-based ANNS algorithms have achieved high search accuracy with practical efficiency. Despite the advancements, these algorithms still face performance bottlenecks in production, due to the random memory access patterns of graph-based search and the high computational overheads of vector distance. In addition, the performance of a graph-based ANNS algorithm is highly sensitive to parameters, while selecting the optimal parameters is cost-prohibitive, e.g., manual tuning requires repeatedly re-building the index. This paper introduces VSAG, an open-source framework that aims to enhance the in production performance of graph-based ANNS algorithms. VSAG has been deployed at scale in the services of Ant Group, and it incorporates three key optimizations: (i) efficient memory access: it reduces L3 cache misses with pre-fetching and cache-friendly vector organization; (ii) automated parameter tuning: it automatically selects performance-optimal parameters without requiring index rebuilding; (iii) efficient distance computation: it leverages modern hardware, scalar quantization, and smartly switches to low-precision representation to dramatically reduce the distance computation costs. We evaluate VSAG on real-world datasets. The experimental results show that VSAG achieves the state-of-the-art performance and provides up to 4x speedup over HNSWlib (an industry-standard library) while ensuring the same accuracy.

cs.DB↗

arXiv2025

ELIS: Efficient LLM Iterative Scheduling System with Response Length Predictor

We propose ELIS, a serving system for Large Language Models (LLMs) featuring an Iterative Shortest Remaining Time First (ISRTF) scheduler designed to efficiently manage inference tasks with the shortest remaining tokens. Current LLM serving systems often employ a first-come-first-served scheduling strategy, which can lead to the "head-of-line blocking" problem. To overcome this limitation, it is necessary to predict LLM inference times and apply a shortest job first scheduling strategy. However, due to the auto-regressive nature of LLMs, predicting the inference latency is challenging. ELIS addresses this challenge by training a response length predictor for LLMs using the BGE model, an encoder-based state-of-the-art model. Additionally, we have devised the ISRTF scheduling strategy, an optimization of shortest remaining time first tailored to existing LLM iteration batching. To evaluate our work in an industrial setting, we simulate streams of requests based on our study of real-world user LLM serving trace records. Furthermore, we implemented ELIS as a cloud-native scheduler system on Kubernetes to evaluate its performance in production environments. Our experimental results demonstrate that ISRTF reduces the average job completion time by up to 19.6%.

cs.DC↗

arXiv2025

ESG: Elastic Graphs for Range-Filtering Approximate k-Nearest Neighbor Search

Range-filtering approximate $k$-nearest neighbor (RFAKNN) search takes as input a vector and a numeric value, returning $k$ points from a database of $N$ high-dimensional points. The returned points must satisfy two criteria: their numeric values must lie within the specified query range, and they must be approximately the $k$ nearest points to the query vector. To strike a better balance between query accuracy and efficiency, we propose novel methods that relax the strict requirement for subranges to \textit{exactly} match the query range. This elastic relaxation is based on a theoretical insight: allowing the controlled inclusion of out-of-range points during the search does not compromise the bounded complexity of the search process. Building on this insight, we prove that our methods reduce the number of required subranges to at most \textit{two}, eliminating the $O(\log N)$ query overhead inherent in existing methods. Extensive experiments on real-world datasets demonstrate that our proposed methods outperform state-of-the-art approaches, achieving performance improvements of 1.5x to 6x while maintaining high accuracy.

cs.DB↗

arXiv2025

Diffusion-Aided Joint Source Channel Coding For High Realism Wireless Image Transmission

Deep learning-based joint source-channel coding (deep JSCC) has been demonstrated to be an effective approach for wireless image transmission. Nevertheless, most existing work adopts an autoencoder framework to optimize conventional criteria such as Mean Squared Error (MSE) and Structural Similarity Index (SSIM) which do not suffice to maintain the perceptual quality of reconstructed images. Such an issue is more prominent under stringent bandwidth constraints or low signal-to-noise ratio (SNR) conditions. To tackle this challenge, we propose DiffJSCC, a novel framework that leverages the prior knowledge of the pre-trained Statble Diffusion model to produce high-realism images via the conditional diffusion denoising process. Our DiffJSCC first extracts multimodal spatial and textual features from the noisy channel symbols in the generation phase. Then, it produces an initial reconstructed image as an intermediate representation to aid robust feature extraction and a stable training process. In the following diffusion step, DiffJSCC uses the derived multimodal features, together with channel state information such as the signal-to-noise ratio (SNR), as conditions to guide the denoising diffusion process, which converts the initial random noise to the final reconstruction. DiffJSCC employs a novel control module to fine-tune the Stable Diffusion model and adjust it to the multimodal conditions. Extensive experiments on diverse datasets reveal that our method significantly surpasses prior deep JSCC approaches on both perceptual metrics and downstream task performance, showcasing its ability to preserve the semantics of the original transmitted images. Notably, DiffJSCC can achieve highly realistic reconstructions for 768x512 pixel Kodak images with only 3072 symbols (<0.008 symbols per pixel) under 1dB SNR channels.

eess.SP↗

arXiv2025

SAM-guided Pseudo Label Enhancement for Multi-modal 3D Semantic Segmentation

Multi-modal 3D semantic segmentation is vital for applications such as autonomous driving and virtual reality (VR). To effectively deploy these models in real-world scenarios, it is essential to employ cross-domain adaptation techniques that bridge the gap between training data and real-world data. Recently, self-training with pseudo-labels has emerged as a predominant method for cross-domain adaptation in multi-modal 3D semantic segmentation. However, generating reliable pseudo-labels necessitates stringent constraints, which often result in sparse pseudo-labels after pruning. This sparsity can potentially hinder performance improvement during the adaptation process. We propose an image-guided pseudo-label enhancement approach that leverages the complementary 2D prior knowledge from the Segment Anything Model (SAM) to introduce more reliable pseudo-labels, thereby boosting domain adaptation performance. Specifically, given a 3D point cloud and the SAM masks from its paired image data, we collect all 3D points covered by each SAM mask that potentially belong to the same object. Then our method refines the pseudo-labels within each SAM mask in two steps. First, we determine the class label for each mask using majority voting and employ various constraints to filter out unreliable mask labels. Next, we introduce Geometry-Aware Progressive Propagation (GAPP) which propagates the mask label to all 3D points within the SAM mask while avoiding outliers caused by 2D-3D misalignment. Experiments conducted across multiple datasets and domain adaptation scenarios demonstrate that our proposed method significantly increases the quantity of high-quality pseudo-labels and enhances the adaptation performance over baseline methods.

cs.CV↗

arXiv2025

Effective and General Distance Computation for Approximate Nearest Neighbor Search

Approximate K Nearest Neighbor (AKNN) search in high-dimensional spaces is a critical yet challenging problem. In AKNN search, distance computation is the core task that dominates the runtime. Existing approaches typically use approximate distances to improve computational efficiency, often at the cost of reduced search accuracy. To address this issue, the state-of-the-art method, ADSampling, employs random projections to estimate approximate distances and introduces an additional distance correction process to mitigate accuracy loss. However, ADSampling has limitations in both effectiveness and generality, primarily due to its reliance on random projections for distance approximation and correction. To address the effectiveness limitations of ADSampling, we leverage data distribution to improve distance computation via orthogonal projection. Furthermore, to overcome the generality limitations of ADSampling, we adopt a data-driven approach to distance correction, decoupling the correction process from the distance approximation process. Extensive experiments demonstrate the superiority and effectiveness of our method. In particular, compared to ADSampling, our method achieves a speedup of 1.6 to 2.1 times on real-world datasets while providing higher accuracy.

cs.DB↗

arXiv2024

Fully and partially distributed Quantum Generalized Benders Decomposition for Unit Commitment Problems

A series of hybrid quantum-classical generalized Benders decomposition (GBD) algorithms are proposed to address unit commitment (UC) problems under centralized, distributed, and partially distributed frameworks. In the centralized approach, the quantum GBD transforms the master problem (MP) into a quadratic unconstrained binary optimization form suitable for quantum computing. For distributed systems, the distributed consensus quantum GBD employs an average consensus strategy to reformulate subproblems into local subproblems. By leveraging the dual information, local cutting planes are constructed to decompose the MP into local master problems (LMPs). This approach reduces the qubit overhead and addresses the partitioning requirements. The consensus-inspired quantum GBD (CIQGBD) and its partially distributed variant, D-CIQGBD are proposed based on optimizing the allocation of relaxation variables directly, the algorithms construct more rational cutting planes, thereby enhancing the minimum eigenenergy gap of the system Hamiltonian during quantum annealing and improving the computational efficiency. Extensive experiments under various UC scenarios validate the performance of the above-mentioned hybrid algorithms. Compared to the classical solver Gurobi, D-CIQGBD demonstrates a speed advantage in solving the security-constrained UC problem on the IEEE-RTS 24-bus system. These results provide new perspectives on leveraging quantum computing for the distributed optimization of power systems.

quant-ph↗

arXiv2024

Playable Game Generation

In recent years, Artificial Intelligence Generated Content (AIGC) has advanced from text-to-image generation to text-to-video and multimodal video synthesis. However, generating playable games presents significant challenges due to the stringent requirements for real-time interaction, high visual quality, and accurate simulation of game mechanics. Existing approaches often fall short, either lacking real-time capabilities or failing to accurately simulate interactive mechanics. To tackle the playability issue, we propose a novel method called \emph{PlayGen}, which encompasses game data generation, an autoregressive DiT-based diffusion model, and a comprehensive playability-based evaluation framework. Validated on well-known 2D and 3D games, PlayGen achieves real-time interaction, ensures sufficient visual quality, and provides accurate interactive mechanics simulation. Notably, these results are sustained even after over 1000 frames of gameplay on an NVIDIA RTX 2060 GPU. Our code is publicly available: https://github.com/GreatX3/Playable-Game-Generation. Our playable demo generated by AI is: http://124.156.151.207.

cs.AI↗

arXiv2024

A Survey of Attacks on Large Vision-Language Models: Resources, Advances, and Future Trends

With the significant development of large models in recent years, Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities across a wide range of multimodal understanding and reasoning tasks. Compared to traditional Large Language Models (LLMs), LVLMs present great potential and challenges due to its closer proximity to the multi-resource real-world applications and the complexity of multi-modal processing. However, the vulnerability of LVLMs is relatively underexplored, posing potential security risks in daily usage. In this paper, we provide a comprehensive review of the various forms of existing LVLM attacks. Specifically, we first introduce the background of attacks targeting LVLMs, including the attack preliminary, attack challenges, and attack resources. Then, we systematically review the development of LVLM attack methods, such as adversarial attacks that manipulate model outputs, jailbreak attacks that exploit model vulnerabilities for unauthorized actions, prompt injection attacks that engineer the prompt type and pattern, and data poisoning that affects model training. Finally, we discuss promising research directions in the future. We believe that our survey provides insights into the current landscape of LVLM vulnerabilities, inspiring more researchers to explore and mitigate potential safety issues in LVLM developments. The latest papers on LVLM attacks are continuously collected in https://github.com/liudaizong/Awesome-LVLM-Attack.

cs.CV↗

arXiv2024

Weak antilocalization and localization in Eu$_2$Ir$_2$O$_7$ (111) thin films by reactive solid phase epitaxy

Thin films of the pyrochlore iridates along the [111] direction have drawn significant attention to investigate exotic correlated topological phenomena. Here, we report the fabrication of Eu$_2$Ir$_2$O$_7$ thin films via reactive solid phase epitaxy using the pulsed laser deposition technique. We mainly focus on the transport properties of the films below the magnetic phase transition at 105 K. Analyses on the temperature and the field dependences of resistivity unveil the presence of weak antilocalization, a characteristic signature of the Weyl semimetallic state that has been "buried" by magnetism. Moreover, it is noteworthy that the contribution from many-body interactions in Eu2Ir2O7 thin films is enhanced at lower temperatures and competes with the weak antilocalization effect, and eventually drives the crossover to weak localization at 2 K.

cond-mat.str-el↗

arXiv2023

Efficient Computation Sharing for Multi-Task Visual Scene Understanding

Solving multiple visual tasks using individual models can be resource-intensive, while multi-task learning can conserve resources by sharing knowledge across different tasks. Despite the benefits of multi-task learning, such techniques can struggle with balancing the loss for each task, leading to potential performance degradation. We present a novel computation- and parameter-sharing framework that balances efficiency and accuracy to perform multiple visual tasks utilizing individually-trained single-task transformers. Our method is motivated by transfer learning schemes to reduce computational and parameter storage costs while maintaining the desired performance. Our approach involves splitting the tasks into a base task and the other sub-tasks, and sharing a significant portion of activations and parameters/weights between the base and sub-tasks to decrease inter-task redundancies and enhance knowledge sharing. The evaluation conducted on NYUD-v2 and PASCAL-context datasets shows that our method is superior to the state-of-the-art transformer-based multi-task learning techniques with higher accuracy and reduced computational resources. Moreover, our method is extended to video stream inputs, further reducing computational costs by efficiently sharing information across the temporal domain as well as the task domain. Our codes and models will be publicly available.

cs.CV↗

arXiv2023

Diving with Penguins: Detecting Penguins and their Prey in Animal-borne Underwater Videos via Deep Learning

African penguins (Spheniscus demersus) are an endangered species. Little is known regarding their underwater hunting strategies and associated predation success rates, yet this is essential for guiding conservation. Modern bio-logging technology has the potential to provide valuable insights, but manually analysing large amounts of data from animal-borne video recorders (AVRs) is time-consuming. In this paper, we publish an animal-borne underwater video dataset of penguins and introduce a ready-to-deploy deep learning system capable of robustly detecting penguins (mAP50@98.0%) and also instances of fish (mAP50@73.3%). We note that the detectors benefit explicitly from air-bubble learning to improve accuracy. Extending this detector towards a dual-stream behaviour recognition network, we also provide the first results for identifying predation behaviour in penguin underwater videos. Whilst results are promising, further work is required for useful applicability of predation behaviour detection in field scenarios. In summary, we provide a highly reliable underwater penguin detector, a fish detector, and a valuable first attempt towards an automated visual detection of complex behaviours in a marine predator. We publish the networks, the DivingWithPenguins video dataset, annotations, splits, and weights for full reproducibility and immediate usability by practitioners.

cs.CV↗