arXiv subjects

Biao Zhang

Publications and source records attributed to Biao Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2025

Probing the strong interaction between charm hadrons and charged particles with femtoscopy measurements with ALICE

Studies of strong interactions between hadrons provide a valuable opportunity to test Quantum Chromodynamics at nucleon-scale distances. The femtoscopy technique has proven to be an effective tool for studying interactions between unstable hadrons by measuring the correlation function of hadron pairs in momentum space. While several measurements of the strong interactions between light and strange hadrons have been conducted using this technique, studies of charm hadrons have been limited. These studies can shed light on the formation of exotic charm states or, in the case of baryons, nuclei with charm content. In these proceedings, measurements of the strong interaction between charm hadrons and light-flavor hadrons using the femtoscopy technique are presented. The final results on the correlation functions and residual strong interactions between $\rm D$ mesons and light hadrons measured in pp collisions at $\sqrt{s} = 13$ TeV are discussed. Additionally, new studies of correlations between $Λ_{\rm c}^+$ and protons in pp collisions at $\sqrt{s} = 13.6$ TeV, utilizing the new data samples collected from LHC Run 3, are presented.

hep-ex↗

arXiv2025

Diamond-based sensing of stray fields from the bulk of thin-film magnets via nano-indentation

Measurement of the magnetization in the bulk of thin-film or two-dimensional materials is important for understanding their intrinsic properties without the complications from edges or domain walls. However, the stray fields from the bulk vanish or are very weak, which limits the application of direct measurement methods. Here, we develop a non-destructive approach to directly measuring the stray fields from the bulk of thin-film magnets at arbitrarily designatable locations, with nanoscale spatial resolution. We employ nano-indentation to induce the leakage of stray fields from the materials and use nano-diamond magnetometers to measure them. We apply the method to iron thin films and determine the intrinsic magnetization in the bulk of the materials. This work provides direct access to the intrinsic magnetic properties of thin-film and low-dimensional materials, as well as a method to study the mechanical effects on magnetization in nanomaterials.

physics.app-ph↗

arXiv2025

EmbeddingGemma: Powerful and Lightweight Text Representations

We introduce EmbeddingGemma, a new lightweight, open text embedding model based on the Gemma 3 language model family. Our innovative training recipe strategically captures knowledge from larger models via encoder-decoder initialization and geometric embedding distillation. We improve model robustness and expressiveness with a spread-out regularizer, and ensure generalizability by merging checkpoints from varied, optimized mixtures. Evaluated on the Massive Text Embedding Benchmark (MTEB) across multilingual, English, and code domains, EmbeddingGemma (300M) achieves state-of-the-art results. Notably, it outperforms prior top models, both proprietary and open, with fewer than 500M parameters, and provides performance comparable to models double its size, offering an exceptional performance-to-cost ratio. Remarkably, this lead persists when quantizing model weights or truncating embedding outputs. This makes EmbeddingGemma particularly well-suited for low-latency and high-throughput use cases such as on-device applications. We provide ablation studies exploring our key design choices. We release EmbeddingGemma to the community to promote further research.

cs.CL↗

arXiv2025

Encoder-Decoder or Decoder-Only? Revisiting Encoder-Decoder Large Language Model

Recent large language model (LLM) research has undergone an architectural shift from encoder-decoder modeling to nowadays the dominant decoder-only modeling. This rapid transition, however, comes without a rigorous comparative analysis especially \textit{from the scaling perspective}, raising concerns that the potential of encoder-decoder models may have been overlooked. To fill this gap, we revisit encoder-decoder LLM (RedLLM), enhancing it with recent recipes from decoder-only LLM (DecLLM). We conduct a comprehensive comparison between RedLLM, pretrained with prefix language modeling (LM), and DecLLM, pretrained with causal LM, at different model scales, ranging from $\sim$150M to $\sim$8B. Using RedPajama V1 (1.6T tokens) for pretraining and FLAN for instruction tuning, our experiments show that RedLLM produces compelling scaling properties and surprisingly strong performance. While DecLLM is overall more compute-optimal during pretraining, RedLLM demonstrates comparable scaling and context length extrapolation capabilities. After instruction tuning, RedLLM achieves comparable and even better results on various downstream tasks while enjoying substantially better inference efficiency. We hope our findings could inspire more efforts on re-examining RedLLM, unlocking its potential for developing powerful and efficient LLMs.

cs.CL↗

arXiv2025

SMEC: Rethinking Matryoshka Representation Learning for Retrieval Embedding Compression

Large language models (LLMs) generate high-dimensional embeddings that capture rich semantic and syntactic information. However, high-dimensional embeddings exacerbate computational complexity and storage requirements, thereby hindering practical deployment. To address these challenges, we propose a novel training framework named Sequential Matryoshka Embedding Compression (SMEC). This framework introduces the Sequential Matryoshka Representation Learning(SMRL) method to mitigate gradient variance during training, the Adaptive Dimension Selection (ADS) module to reduce information degradation during dimension pruning, and the Selectable Cross-batch Memory (S-XBM) module to enhance unsupervised learning between high- and low-dimensional embeddings. Experiments on image, text, and multimodal datasets demonstrate that SMEC achieves significant dimensionality reduction while maintaining performance. For instance, on the BEIR dataset, our approach improves the performance of compressed LLM2Vec embeddings (256 dimensions) by 1.1 points and 2.7 points compared to the Matryoshka-Adaptor and Search-Adaptor models, respectively.

cs.CL↗

arXiv2025

Meaningful Pose-Based Sign Language Evaluation

We present a comprehensive study on meaningfully evaluating sign language utterances in the form of human skeletal poses. The study covers keypoint distance-based, embedding-based, and back-translation-based metrics. We show tradeoffs between different metrics in different scenarios through automatic meta-evaluation of sign-level retrieval and a human correlation study of text-to-pose translation across different sign languages. Our findings and the open-source pose-evaluation toolkit provide a practical and reproducible way of developing and evaluating sign language translation or generation systems.

cs.CL↗

arXiv2025

MatCLIP: Light- and Shape-Insensitive Assignment of PBR Material Models

Assigning realistic materials to 3D models remains a significant challenge in computer graphics. We propose MatCLIP, a novel method that extracts shape- and lighting-insensitive descriptors of Physically Based Rendering (PBR) materials to assign plausible textures to 3D objects based on images, such as the output of Latent Diffusion Models (LDMs) or photographs. Matching PBR materials to static images is challenging because the PBR representation captures the dynamic appearance of materials under varying viewing angles, shapes, and lighting conditions. By extending an Alpha-CLIP-based model on material renderings across diverse shapes and lighting, and encoding multiple viewing conditions for PBR materials, our approach generates descriptors that bridge the domains of PBR representations with photographs or renderings, including LDM outputs. This enables consistent material assignments without requiring explicit knowledge of material relationships between different parts of an object. MatCLIP achieves a top-1 classification accuracy of 76.6%, outperforming state-of-the-art methods such as PhotoShape and MatAtlas by over 15 percentage points on publicly available datasets. Our method can be used to construct material assignments for 3D shape datasets such as ShapeNet, 3DCoMPaT++, and Objaverse. All code and data will be released.

cs.CV↗

arXiv2025

V2M4: 4D Mesh Animation Reconstruction from a Single Monocular Video

We present V2M4, a novel 4D reconstruction method that directly generates a usable 4D mesh animation asset from a single monocular video. Unlike existing approaches that rely on priors from multi-view image and video generation models, our method is based on native 3D mesh generation models. Naively applying 3D mesh generation models to generate a mesh for each frame in a 4D task can lead to issues such as incorrect mesh poses, misalignment of mesh appearance, and inconsistencies in mesh geometry and texture maps. To address these problems, we propose a structured workflow that includes camera search and mesh reposing, condition embedding optimization for mesh appearance refinement, pairwise mesh registration for topology consistency, and global texture map optimization for texture consistency. Our method outputs high-quality 4D animated assets that are compatible with mainstream graphics and game software. Experimental results across a variety of animation types and motion amplitudes demonstrate the generalization and effectiveness of our method. Project page: https://windvchen.github.io/V2M4/.

cs.GR↗

arXiv2025

efunc: An Efficient Function Representation without Neural Networks

Function fitting/approximation plays a fundamental role in computer graphics and other engineering applications. While recent advances have explored neural networks to address this task, these methods often rely on architectures with many parameters, limiting their practical applicability. In contrast, we pursue high-quality function approximation using parameter-efficient representations that eliminate the dependency on neural networks entirely. We first propose a novel framework for continuous function modeling. Most existing works can be formulated using this framework. We then introduce a compact function representation, which is based on polynomials interpolated using radial basis functions, bypassing both neural networks and complex/hierarchical data structures. We also develop memory-efficient CUDA-optimized algorithms that reduce computational time and memory consumption to less than 10% compared to conventional automatic differentiation frameworks. Finally, we validate our representation and optimization pipeline through extensive experiments on 3D signed distance functions (SDFs). The proposed representation achieves comparable or superior performance to state-of-the-art techniques (e.g., octree/hash-grid techniques) with significantly fewer parameters.

cs.GR↗

arXiv2025

Hydrogen-poor Superluminous Supernovae with Bumpy Light Curves Powered by Precessing Magnetars

Recent observations and statistical studies have revealed that a significant fraction of hydrogen-poor superluminous supernovae (SLSNe-I) exhibit light curves that deviate from the smooth evolution predicted by the magnetar-powered model, instead showing one or more bumps after the primary peak. However, the formation mechanisms of these post-peak bumps remain a matter of debate. Furthermore, previous studies employing the magnetar-powered model have typically assumed a fixed magnetic inclination angle and neglected the effects of magnetar precession. However, recent research has shown that the precession of newborn magnetars forming during the collapse of massive stars causes the magnetic inclination angle to evolve over time, thereby influencing magnetic dipole radiation. In this paper, therefore, we incorporate the effects of magnetar precession into the magnetar-powered model to develop the precessing magnetar-powered model. Using this model, we successfully reproduce the multi-band light curves of 6 selected representative SLSNe-I with post-peak bumps. Moreover, the derived model parameters fall within the typical parameter range for SLSNe-I. By combining the precessing magnetars in SLSNe-I and long GRBs, we find that the ellipticity of magnetars is related to the dipole magnetic field strength, which may suggest a common origin for the two phenomena. Our work provides a potential explanation for the origin of post-peak bumps in SLSNe-I and offers evidence for the early precession of newborn magnetars formed in supernova explosions.

astro-ph.HE↗

arXiv2025

Encoder-Decoder Gemma: Improving the Quality-Efficiency Trade-Off via Adaptation

While decoder-only large language models (LLMs) have shown impressive results, encoder-decoder models are still widely adopted in real-world applications for their inference efficiency and richer encoder representation. In this paper, we study a novel problem: adapting pretrained decoder-only LLMs to encoder-decoder, with the goal of leveraging the strengths of both approaches to achieve a more favorable quality-efficiency trade-off. We argue that adaptation not only enables inheriting the capability of decoder-only LLMs but also reduces the demand for computation compared to pretraining from scratch. We rigorously explore different pretraining objectives and parameter initialization/optimization techniques. Through extensive experiments based on Gemma 2 (2B and 9B) and a suite of newly pretrained mT5-sized models (up to 1.6B), we demonstrate the effectiveness of adaptation and the advantage of encoder-decoder LLMs. Under similar inference budget, encoder-decoder LLMs achieve comparable (often better) pretraining performance but substantially better finetuning performance than their decoder-only counterpart. For example, Gemma 2B-2B outperforms Gemma 2B by $\sim$7\% after instruction tuning. Encoder-decoder adaptation also allows for flexible combination of different-sized models, where Gemma 9B-2B significantly surpasses Gemma 2B-2B by $>$3\%. The adapted encoder representation also yields better results on SuperGLUE. We will release our checkpoints to facilitate future research.

cs.CL↗

arXiv2025

iFlame: Interleaving Full and Linear Attention for Efficient Mesh Generation

This paper propose iFlame, a novel transformer-based network architecture for mesh generation. While attention-based models have demonstrated remarkable performance in mesh generation, their quadratic computational complexity limits scalability, particularly for high-resolution 3D data. Conversely, linear attention mechanisms offer lower computational costs but often struggle to capture long-range dependencies, resulting in suboptimal outcomes. To address this trade-off, we propose an interleaving autoregressive mesh generation framework that combines the efficiency of linear attention with the expressive power of full attention mechanisms. To further enhance efficiency and leverage the inherent structure of mesh representations, we integrate this interleaving approach into an hourglass architecture, which significantly boosts efficiency. Our approach reduces training time while achieving performance comparable to pure attention-based models. To improve inference efficiency, we implemented a caching algorithm that almost doubles the speed and reduces the KV cache size by seven-eighths compared to the original Transformer. We evaluate our framework on ShapeNet and Objaverse, demonstrating its ability to generate high-quality 3D meshes efficiently. Our results indicate that the proposed interleaving framework effectively balances computational efficiency and generative performance, making it a practical solution for mesh generation. The training takes only 2 days with 4 GPUs on 39k data with a maximum of 4k faces on Objaverse.

cs.CV↗

arXiv2025

WMT24++: Expanding the Language Coverage of WMT24 to 55 Languages & Dialects

As large language models (LLM) become more and more capable in languages other than English, it is important to collect benchmark datasets in order to evaluate their multilingual performance, including on tasks like machine translation (MT). In this work, we extend the WMT24 dataset to cover 55 languages by collecting new human-written references and post-edits for 46 new languages and dialects in addition to post-edits of the references in 8 out of 9 languages in the original WMT24 dataset. The dataset covers four domains: literary, news, social, and speech. We benchmark a variety of MT providers and LLMs on the collected dataset using automatic metrics and find that LLMs are the best-performing MT systems in all 55 languages. These results should be confirmed using a human-based evaluation, which we leave for future work.

cs.CL↗

arXiv2025

Autoregressive Generation of Static and Growing Trees

We propose a transformer architecture and training strategy for tree generation. The architecture processes data at multiple resolutions and has an hourglass shape, with middle layers processing fewer tokens than outer layers. Similar to convolutional networks, we introduce longer range skip connections to completent this multi-resolution approach. The key advantage of this architecture is the faster processing speed and lower memory consumption. We are therefore able to process more complex trees than would be possible with a vanilla transformer architecture. Furthermore, we extend this approach to perform image-to-tree and point-cloud-to-tree conditional generation and to simulate the tree growth processes, generating 4D trees. Empirical results validate our approach in terms of speed, memory consumption, and generation quality.

cs.CV↗

arXiv2024

Signature of Triaxially Precessing Magnetars in Gamma-ray Burst X-Ray Afterglows

The X-ray afterglows of some gamma-ray bursts (GRBs) exhibit plateaus, which can be explained by the internal dissipation of a newborn millisecond magnetar wind. In the early phase of these newborn magnetars, the magnetic inclination angle undergoes periodic changes due to precession, leading to periodic modulation of the injection luminosity due to magnetic dipole radiation. This may result in quasi-periodic oscillations (QPOs) on the plateaus. In this paper, we identify four GRBs with regular flux variations on their X-ray afterglow plateaus from Swift/XRT data before November 2023, three of which exhibit periodicity. Based on the likelihood of supporting a precessing magnetar as the central engine, we classify them into three categories: Gold (GRB 060202 and GRB 180620A), Silver (GRB 050730), and Bronze (GRB 210610A). We invoke a model of magnetic dipole radiation emitted by a triaxially freely precessing magnetar whose spin-down is dominated by electromagnetic radiation, to fit the light curves. Our model successfully reproduces the light curves of these four GRBs, including the regular flux variations on the plateaus and their periodicity (if present). Our work provides further evidence for early precession in newborn millisecond magnetars in GRBs.

astro-ph.HE↗

arXiv2024

LLMRefine: Pinpointing and Refining Large Language Models via Fine-Grained Actionable Feedback

Recent large language models (LLM) are leveraging human feedback to improve their generation quality. However, human feedback is costly to obtain, especially during inference. In this work, we propose LLMRefine, an inference time optimization method to refine LLM's output. The core idea is to use a learned fine-grained feedback model to pinpoint defects and guide LLM to refine them iteratively. Using original LLM as a proposal of edits, LLMRefine searches for defect-less text via simulated annealing, trading off the exploration and exploitation. We conduct experiments on three text generation tasks, including machine translation, long-form question answering (QA), and topical summarization. LLMRefine consistently outperforms all baseline approaches, achieving improvements up to 1.7 MetricX points on translation tasks, 8.1 ROUGE-L on ASQA, 2.2 ROUGE-L on topical summarization.

cs.CL↗

arXiv2024

Many-Shot In-Context Learning

Large language models (LLMs) excel at few-shot in-context learning (ICL) -- learning from a few examples provided in context at inference, without any weight updates. Newly expanded context windows allow us to investigate ICL with hundreds or thousands of examples -- the many-shot regime. Going from few-shot to many-shot, we observe significant performance gains across a wide variety of generative and discriminative tasks. While promising, many-shot ICL can be bottlenecked by the available amount of human-generated examples. To mitigate this limitation, we explore two new settings: Reinforced and Unsupervised ICL. Reinforced ICL uses model-generated chain-of-thought rationales in place of human examples. Unsupervised ICL removes rationales from the prompt altogether, and prompts the model only with domain-specific questions. We find that both Reinforced and Unsupervised ICL can be quite effective in the many-shot regime, particularly on complex reasoning tasks. Finally, we demonstrate that, unlike few-shot learning, many-shot learning is effective at overriding pretraining biases, can learn high-dimensional functions with numerical inputs, and performs comparably to fine-tuning. We also find that inference cost increases linearly in the many-shot regime, and frontier LLMs benefit from many-shot ICL to varying degrees. Our analysis also reveals the limitations of next-token prediction loss as an indicator of downstream ICL performance.

cs.LG↗

arXiv2024

Identifying the Origin of FRB-associated X-ray Bursts with X-ray Polarization

The origin of extraordinary X-ray burst (XRB) associated with a fast radio burst (FRB) like FRB 20200428D is still unclear, though several models such as the emission of a trapped fireball modified by resonant cyclotron scattering, the outflow from a polar trapped-expanding fireball, and the synchrotron radiation of a far-away relativistic shock, have been proposed. To determine which model is true, we study possible X-ray polarization signature for each model, inspired by the importance of radio polarization in identifying FRB origin. We first numerically simulate or calculate the XRB spectrum for each model and fit it to the observed data, then compute the corresponding polarization signal based on the fit. We find that these three models predict different polarization patterns in terms of phase/time and energy variations. The differences can be used to test the models with future X-ray polarization observations.

astro-ph.HE↗