Search

Large language models have demonstrated substantial advancements in reasoning capabilities, particularly through inference-time scaling, as illustrated by models such as OpenAI's o1. However, current Vision-Language Models (VLMs) often struggle to perform systematic and structured reasoning, especially when handling complex visual question-answering tasks. In this work, we introduce LLaVA-o1, a novel VLM designed to conduct autonomous multistage reasoning. Unlike chain-of-thought prompting, LLaVA-o1 independently engages in sequential stages of summarization, visual interpretation, logical reasoning, and conclusion generation. This structured approach enables LLaVA-o1 to achieve marked improvements in precision on reasoning-intensive tasks. To accomplish this, we compile the LLaVA-o1-100k dataset, integrating samples from various visual question answering sources and providing structured reasoning annotations. Besides, we propose an inference-time stage-level beam search method, which enables effective inference-time scaling. Remarkably, with only 100k training samples and a simple yet effective inference time scaling method, LLaVA-o1 not only outperforms its base model by 8.9% on a wide range of multimodal reasoning benchmarks, but also surpasses the performance of larger and even closed-source models, such as Gemini-1.5-pro, GPT-4o-mini, and Llama-3.2-90B-Vision-Instruct., Comment: 11 pages, 5 figures

Modeling and simulating the protein folding process overall remains a grand challenge in computational biology. We systematically investigate end-to-end quantum algorithms for simulating various protein dynamics with effects, such as mechanical forces or stochastic noises. We offer efficient quantum simulation algorithms to produce quantum encoding of the final states, history states, or density matrices of inhomogeneous or stochastic harmonic oscillator models. For the read-in setting, we design (i) efficient quantum algorithms for initial state preparation, utilizing counter-based random number generator and rejection sampling, and (ii) depth-efficient approaches for molecular structure loading. Both are particularly important in handling large protein molecules. For the read-out setting, our algorithms estimate various classical observables, such as energy, low vibration modes, density of states, correlation of displacement, and optimal control of molecular dynamics. We also show classical numerical experiments focused on estimating the density of states and applying the optimal control to facilitate conformation changes to verify our arguments on potential quantum speedups. Overall, our study demonstrates that the quantum simulation of protein dynamics can be a solid end-to-end application in the era of early or fully fault-tolerant quantum computing., Comment: 61 pages, 10 figures

In this work, we upgrade the multi-head attention mechanism, the core of the Transformer model, to improve efficiency while maintaining or surpassing the previous accuracy level. We show that multi-head attention can be expressed in the summation form. Drawing on the insight that not all attention heads hold equal significance, we propose Mixture-of-Head attention (MoH), a new architecture that treats attention heads as experts in the Mixture-of-Experts (MoE) mechanism. MoH has two significant advantages: First, MoH enables each token to select the appropriate attention heads, enhancing inference efficiency without compromising accuracy or increasing the number of parameters. Second, MoH replaces the standard summation in multi-head attention with a weighted summation, introducing flexibility to the attention mechanism and unlocking extra performance potential. Extensive experiments on ViT, DiT, and LLMs demonstrate that MoH outperforms multi-head attention by using only 50%-90% of the attention heads. Moreover, we demonstrate that pre-trained multi-head attention models, such as LLaMA3-8B, can be further continue-tuned into our MoH models. Notably, MoH-LLaMA3-8B achieves an average accuracy of 64.0% across 14 benchmarks, outperforming LLaMA3-8B by 2.4% by utilizing only 75% of the attention heads. We believe the proposed MoH is a promising alternative to multi-head attention and provides a strong foundation for developing advanced and efficient attention-based models., Comment: 23 pages, code: https://github.com/SkyworkAI/MoH

In this work, we aim to simultaneously enhance the effectiveness and efficiency of Mixture-of-Experts (MoE) methods. To achieve this, we propose MoE++, a general and heterogeneous MoE framework that integrates both Feed-Forward Network~(FFN) and zero-computation experts. Specifically, we introduce three types of zero-computation experts: the zero expert, copy expert, and constant expert, which correspond to discard, skip, and replace operations, respectively. This design offers three key advantages: (i) Low Computing Overhead: Unlike the uniform mixing mechanism for all tokens within vanilla MoE, MoE++ allows each token to engage with a dynamic number of FFNs, be adjusted by constant vectors, or even skip the MoE layer entirely. (ii) High Performance: By enabling simple tokens to utilize fewer FFN experts, MoE++ allows more experts to focus on challenging tokens, thereby unlocking greater performance potential than vanilla MoE. (iii) Deployment Friendly: Given that zero-computation experts have negligible parameters, we can deploy all zero-computation experts on each GPU, eliminating the significant communication overhead and expert load imbalance associated with FFN experts distributed across different GPUs. Moreover, we leverage gating residuals, enabling each token to consider the pathway taken in the previous layer when selecting the appropriate experts. Extensive experimental results demonstrate that MoE++ achieves better performance while delivering 1.1-2.1x expert forward throughput compared to a vanilla MoE model of the same size, which lays a solid foundation for developing advanced and efficient MoE-related models., Comment: 23 pages, Code: https://github.com/SkyworkAI/MoE-plus-plus

Thermal transport is a fundamental mechanism of energy transfer process quite distinct from wave propagation phenomena. It can be manipulated well beyond the possibilities offered by natural materials with a new generation of artificial metamaterials: thermal metamaterials. Topological physics, a focal point in contemporary condensed matter physics, is closely intertwined with thermal metamaterials in recent years. Inspired by topological photonics and topological acoustics in wave metamaterials, a new research field emerged recently, which we dub `topological thermotics', which encompasses three primary branches: topological thermal conduction, convection, and radiation. For topological thermal conduction, we discuss recent advances in both 1D and higher-dimensional thermal topological phases. For topological thermal convection, we discuss the implementation of thermal exceptional points with their unique properties and non-Hermitian thermal topological states. Finally, we review the most recent demonstration of topological effects in the near-field and far-field radiation. Anticipating future developments, we conclude by discussing potential directions of topological thermotics, including the expansion into other diffusion processes such as particle dynamics and plasma physics, and the integration with machine learning techniques., Comment: This perpective summarizes the topological physics in thermal metamaterials and proposes a new research field, "topological thermotics"

A non-extensive (3+1)-dimensional hydrodynamic model for multi-particle production processes, NEX-CLVisc, is developed in the framework of CLVisc where the viscous corrections are turned off. It assumes that the non-extensive effects consistently exist in the initial conditions set by the optical Glauber model, the equation of state and the hadron kinetic freeze-out procedure. The model is then applied to simulate the pseudo-rapidity ($\eta$) distribution, the transverse momentum ($p_{\rm T}$) spectra and the $p_{\rm T}$-differential elliptic flow ($v_2$) of charged particles in Pb-Pb collisions at $\sqrt{s_{\rm NN}}=$ 2.76 TeV and 5.02 TeV, respectively. It is found that the model can reasonably well reproduce the experimental data of the $\eta$ distribution and the charged-particle spectra in a $p_{\rm T}$ range up to 6-8 GeV/c. When compared with the ideal hydrodynamic model, the $p_{\rm T}$-differential $v_2$ of charged particles is suppressed in the NEX-CLVisc model, which is similar to that observed in the hydrodynamic model with a shear viscous correction. Moreover, due to the lack of the viscous corrections and the event-by-event fluctuation, the model can only describe the $p_{\rm T}$-differential $v_2$ up to 3-4 GeV/c, which is smaller than its applicable range for the particle $p_{\rm T}$ spectra., Comment: 9 pages, 4 figures, 4 tables. Revised accroding to the referee's comments

Text-Video Retrieval (TVR) aims to align and associate relevant video content with corresponding natural language queries. Most existing TVR methods are based on large-scale pre-trained vision-language models (e.g., CLIP). However, due to the inherent plain structure of CLIP, few TVR methods explore the multi-scale representations which offer richer contextual information for a more thorough understanding. To this end, we propose MUSE, a multi-scale mamba with linear computational complexity for efficient cross-resolution modeling. Specifically, the multi-scale representations are generated by applying a feature pyramid on the last single-scale feature map. Then, we employ the Mamba structure as an efficient multi-scale learner to jointly learn scale-wise representations. Furthermore, we conduct comprehensive studies to investigate different model structures and designs. Extensive results on three popular benchmarks have validated the superiority of MUSE., Comment: 8 pages

Inspired by the BESIII newest observation of X(2370) glueball-like particle, we search its productions in both $e^+e^-$ collisions at $\sqrt{s}=$ 4.95 GeV and proton-proton (pp) collisions at $\sqrt{s}=$ 13 TeV with a parton and hadron cascade model PACIAE. In this model, the final partonic state (FPS) and the final hadronic state (FHS) are consecutively simulated and recorded. The X(2370) glueball- or tetraquark-state is then, respectively, recombined by two gluons or four quarks $ss\bar{s}\bar{s}$ in the FPS using the quantum statistical mechanics inspired dynamically constrained phase-space coalescence (DCPC) model. The X(2370) molecular-state is recombined by the baryon-antibaryon of $\Lambda$-$\bar{\Lambda}$ or $\Sigma$-$\bar{\Sigma}$, or by three mesons of $\pi^+\pi^{-}\eta'$, $K^+K^-\eta'$, or $K_S^0K_S^0\eta'$ in the FHS using DCPC model. In both $e^+e^-$ and pp collisions, significant discrepancies in the yields, the transverse momentum spectra and the rapidity distributions among the X(2370) glueball-, tetraquark-, and molecular-state are observed. These discrepancies are proposed as valuable criteria identifying the X(2370) different states from each other. Our results not only support the BESIII observation of glueball-like particle $\rm X(2370)$ production in $e^+e^-$ collisions, but also serve as a prediction for the $\rm X(2370)$ production in pp collisions. We strongly suggest the experimental measurement of the X(2370) glueball-like particle production in pp collisions at the LHC energies., Comment: 6 pages, 5 figures

Quantum-classical hybrid dynamics is crucial for accurately simulating complex systems where both quantum and classical behaviors need to be considered. However, coupling between classical and quantum degrees of freedom and the exponential growth of the Hilbert space present significant challenges. Current machine learning approaches for predicting such dynamics, while promising, remain unknown in their error bounds, sample complexity, and generalizability. In this work, we establish a generic theoretical framework for analyzing quantum-classical adiabatic dynamics with learning algorithms. Based on quantum information theory, we develop a provably efficient adiabatic learning (PEAL) algorithm with logarithmic system size sampling complexity and favorable time scaling properties. We benchmark PEAL on the Holstein model, and demonstrate its accuracy in predicting single-path dynamics and ensemble dynamics observables as well as transfer learning over a family of Hamiltonians. Our framework and algorithm open up new avenues for reliable and efficient learning of quantum-classical dynamics.

In this work, we introduce Gemma 2, a new addition to the Gemma family of lightweight, state-of-the-art open models, ranging in scale from 2 billion to 27 billion parameters. In this new version, we apply several known technical modifications to the Transformer architecture, such as interleaving local-global attentions (Beltagy et al., 2020a) and group-query attention (Ainslie et al., 2023). We also train the 2B and 9B models with knowledge distillation (Hinton et al., 2015) instead of next token prediction. The resulting models deliver the best performance for their size, and even offer competitive alternatives to models that are 2-3 times bigger. We release all our models to the community.

Simulation of physical systems is one of the most promising use cases of future digital quantum computers. In this work we systematically analyze the quantum circuit complexities of block encoding the discretized elliptic operators that arise extensively in numerical simulations for partial differential equations, including high-dimensional instances for many-body simulations. When restricted to rectangular domains with separable boundary conditions, we provide explicit circuits to block encode the many-body Laplacian with separable periodic, Dirichlet, Neumann, and Robin boundary conditions, using standard discretization techniques from low-order finite difference methods. To obtain high-precision, we introduce a scheme based on periodic extensions to solve Dirichlet and Neumann boundary value problems using a high-order finite difference method, with only a constant increase in total circuit depth and subnormalization factor. We then present a scheme to implement block encodings of differential operators acting on more arbitrary domains, inspired by Cartesian immersed boundary methods. We then block encode the many-body convective operator, which describes interacting particles experiencing a force generated by a pair-wise potential given as an inverse power law of the interparticle distance. This work provides concrete recipes that are readily translated into quantum circuits, with depth logarithmic in the total Hilbert space dimension, that block encode operators arising broadly in applications involving the quantum simulation of quantum and classical many-body mechanics.

Text-to-motion generation requires not only grounding local actions in language but also seamlessly blending these individual actions to synthesize diverse and realistic global motions. However, existing motion generation methods primarily focus on the direct synthesis of global motions while neglecting the importance of generating and controlling local actions. In this paper, we propose the local action-guided motion diffusion model, which facilitates global motion generation by utilizing local actions as fine-grained control signals. Specifically, we provide an automated method for reference local action sampling and leverage graph attention networks to assess the guiding weight of each local action in the overall motion synthesis. During the diffusion process for synthesizing global motion, we calculate the local-action gradient to provide conditional guidance. This local-to-global paradigm reduces the complexity associated with direct global motion generation and promotes motion diversity via sampling diverse actions as conditions. Extensive experiments on two human motion datasets, i.e., HumanML3D and KIT, demonstrate the effectiveness of our method. Furthermore, our method provides flexibility in seamlessly combining various local actions and continuous guiding weight adjustment, accommodating diverse user preferences, which may hold potential significance for the community. The project page is available at https://jpthu17.github.io/GuidedMotion-project/., Comment: Accepted by ECCV 2024

Developing prompt-based methods with Large Language Models (LLMs) requires making numerous decisions, which give rise to a combinatorial search problem. For example, selecting the right pre-trained LLM, prompt, and hyperparameters to attain the best performance for a task typically necessitates evaluating an expoential number of candidates on large validation sets. This exhaustive evaluation can be time-consuming and costly, as both inference and evaluation of LLM-based approaches are resource-intensive. Worse, a lot of computation is wasted: Many hyper-parameter settings are non-competitive, and many samples from the validation set are highly correlated - providing little or no new information. So, if the goal is to identify the best method, it can be done far more efficiently if the validation samples and methods are selected adaptively. In this paper, we propose a novel method to address this challenge. We lean on low-rank matrix factorization to fill in missing evaluations and on multi-armed bandits to sequentially identify the next (method, validation sample)-pair to evaluate. We carefully assess the efficacy of our approach on several competitive benchmark problems and show that it can identify the top-performing method using only 5-15% of the typically needed resources -- resulting in a staggering 85-95% LLM cost savings.

This paper presents a novel approach to aligning large language models (LLMs) with individual human preferences, sometimes referred to as Reinforcement Learning from \textit{Personalized} Human Feedback (RLPHF). Given stated preferences along multiple dimensions, such as helpfulness, conciseness, or humor, the goal is to create an LLM without re-training that best adheres to this specification. Starting from specialized expert LLMs, each trained for one such particular preference dimension, we propose a black-box method that merges their outputs on a per-token level. We train a lightweight Preference Control Model (PCM) that dynamically translates the preference description and current context into next-token prediction weights. By combining the expert models' outputs at the token level, our approach dynamically generates text that optimizes the given preference. Empirical tests show that our method matches or surpasses existing preference merging techniques, providing a scalable, efficient alternative to fine-tuning LLMs for individual personalization.

MetaDesigner revolutionizes artistic typography synthesis by leveraging the strengths of Large Language Models (LLMs) to drive a design paradigm centered around user engagement. At the core of this framework lies a multi-agent system comprising the Pipeline, Glyph, and Texture agents, which collectively enable the creation of customized WordArt, ranging from semantic enhancements to the imposition of complex textures. MetaDesigner incorporates a comprehensive feedback mechanism that harnesses insights from multimodal models and user evaluations to refine and enhance the design process iteratively. Through this feedback loop, the system adeptly tunes hyperparameters to align with user-defined stylistic and thematic preferences, generating WordArt that not only meets but exceeds user expectations of visual appeal and contextual relevance. Empirical validations highlight MetaDesigner's capability to effectively serve diverse WordArt applications, consistently producing aesthetically appealing and context-sensitive results., Comment: 18 pages, 16 figures, Project: https://modelscope.cn/studios/WordArt/WordArt

Long-context Multimodal Large Language Models (MLLMs) demand substantial computational resources for inference as the growth of their multimodal Key-Value (KV) cache, in response to increasing input lengths, challenges memory and time efficiency. Unlike single-modality LLMs that manage only textual contexts, the KV cache of long-context MLLMs includes representations from multiple images with temporal and spatial relationships and related textual contexts. The predominance of image tokens means traditional optimizations for LLMs' KV caches are unsuitable for multimodal long-context settings, and no prior works have addressed this challenge. In this work, we introduce LOOK-M, a pioneering, fine-tuning-free approach that efficiently reduces the multimodal KV cache size while maintaining performance comparable to a full cache. We observe that during prompt prefill, the model prioritizes more textual attention over image features, and based on the multimodal interaction observation, a new proposed text-prior method is explored to compress the KV cache. Furthermore, to mitigate the degradation of image contextual information, we propose several compensatory strategies using KV pairs merging. LOOK-M demonstrates that with a significant reduction in KV Cache memory usage, such as reducing it by 80% in some cases, it not only achieves up to 1.5x faster decoding but also maintains or even enhances performance across a variety of long context multimodal tasks.

Text-Video Retrieval (TVR) aims to align relevant video content with natural language queries. To date, most state-of-the-art TVR methods learn image-to-video transfer learning based on large-scale pre-trained visionlanguage models (e.g., CLIP). However, fully fine-tuning these pre-trained models for TVR incurs prohibitively expensive computation costs. To this end, we propose to conduct efficient text-video Retrieval with a sparse-andcorrelated AdaPter (RAP), i.e., fine-tuning the pre-trained model with a few parameterized layers. To accommodate the text-video scenario, we equip our RAP with two indispensable characteristics: temporal sparsity and correlation. Specifically, we propose a low-rank modulation module to refine the per-image features from the frozen CLIP backbone, which accentuates salient frames within the video features while alleviating temporal redundancy. Besides, we introduce an asynchronous self-attention mechanism that first selects the top responsive visual patches and augments the correlation modeling between them with learnable temporal and patch offsets. Extensive experiments on four TVR datasets demonstrate that RAP achieves superior or comparable performance compared to the fully fine-tuned counterpart and other parameter-efficient fine-tuning methods., Comment: Accepted by ACL 2024 Findings

Iteratively improving and repairing source code with large language models (LLMs), known as refinement, has emerged as a popular way of generating programs that would be too complex to construct in one shot. Given a bank of test cases, together with a candidate program, an LLM can improve that program by being prompted with failed test cases. But it remains an open question how to best iteratively refine code, with prior work employing simple greedy or breadth-first strategies. We show here that refinement exposes an explore-exploit tradeoff: exploit by refining the program that passes the most test cases, or explore by refining a lesser considered program. We frame this as an arm-acquiring bandit problem, which we solve with Thompson Sampling. The resulting LLM-based program synthesis algorithm is broadly applicable: Across loop invariant synthesis, visual reasoning puzzles, and competition programming problems, we find that our new method can solve more problems using fewer language model calls.

The imminent era of error-corrected quantum computing urgently demands robust methods to characterize complex quantum states, even from limited and noisy measurements. We introduce the Quantum Attention Network (QuAN), a versatile classical AI framework leveraging the power of attention mechanisms specifically tailored to address the unique challenges of learning quantum complexity. Inspired by large language models, QuAN treats measurement snapshots as tokens while respecting their permutation invariance. Combined with a novel parameter-efficient mini-set self-attention block (MSSAB), such data structure enables QuAN to access high-order moments of the bit-string distribution and preferentially attend to less noisy snapshots. We rigorously test QuAN across three distinct quantum simulation settings: driven hard-core Bose-Hubbard model, random quantum circuits, and the toric code under coherent and incoherent noise. QuAN directly learns the growth in entanglement and state complexity from experimentally obtained computational basis measurements. In particular, it learns the growth in complexity of random circuit data upon increasing depth from noisy experimental data. Taken to a regime inaccessible by existing theory, QuAN unveils the complete phase diagram for noisy toric code data as a function of both noise types. This breakthrough highlights the transformative potential of using purposefully designed AI-driven solutions to assist quantum hardware.

In this paper, 2-inch free-standing diamonds were prepared by using heteroepitaxy on composite Ir/YSZ/Si (001) substrates. To release stress, patterned templates were fabricated using laser etching after the initial growth of 50-nm-diamond. Then, the subsequent growth was completed on a patterned template. The full width at half maximum of the diamond (400) and (311) X-ray rocking curves were 313.5 and 359.3 arcsecs, respectively. Strong band-edge emission in the cathodoluminescence spectrum of the resulting diamond revealed excellent crystalline quality. Furthermore, the 2D mapping of Raman spectra was conducted on a $2 mm \times 2 mm$ area located at the center of the 2-inch sample with a thickness of $400 {\mu}m$. The result showed an average peak width of $2.85 \pm 0.36 cm^{-1}$ and residual stress of $-0.03 \pm 0.37 GPa$. The dislocation density, determined by counting etching pits generated from $ H_2/O_2$ plasma etching, was estimated to be around $2.2 \times 10^7 cm^{-2}$. These results evidence that the laser-patterned method can effectively release stress during the growth of large-size diamonds, offering a simpler and more cost-effective alternative to the traditional photolithography-patterned scheme., Comment: 13 pages, 5 figures

Large language models (LLM), such as Google's Minerva and OpenAI's GPT families, are becoming increasingly capable of solving mathematical quantitative reasoning problems. However, they still make unjustified logical and computational errors in their reasoning steps and answers. In this paper, we leverage the fact that if the training corpus of LLMs contained sufficiently many examples of formal mathematics (e.g. in Isabelle, a formal theorem proving environment), they can be prompted to translate i.e. autoformalize informal mathematical statements into formal Isabelle code -- which can be verified automatically for internal consistency. This provides a mechanism to automatically reject solutions whose formalized versions are inconsistent within themselves or with the formalized problem statement. We evaluate our method on GSM8K, MATH and MultiArith datasets and demonstrate that our approach provides a consistently better heuristic than vanilla majority voting -- the previously best method to identify correct answers, by more than 12% on GSM8K. In our experiments it improves results consistently across all datasets and LLM model sizes. The code can be found at https://github.com/jinpz/dtv., Comment: ICLR 2024

Multimodal Large Language Model (MLLMs) leverages Large Language Models as a cognitive framework for diverse visual-language tasks. Recent efforts have been made to equip MLLMs with visual perceiving and grounding capabilities. However, there still remains a gap in providing fine-grained pixel-level perceptions and extending interactions beyond text-specific inputs. In this work, we propose {\bf{AnyRef}}, a general MLLM model that can generate pixel-wise object perceptions and natural language descriptions from multi-modality references, such as texts, boxes, images, or audio. This innovation empowers users with greater flexibility to engage with the model beyond textual and regional prompts, without modality-specific designs. Through our proposed refocusing mechanism, the generated grounding output is guided to better focus on the referenced object, implicitly incorporating additional pixel-level supervision. This simple modification utilizes attention scores generated during the inference of LLM, eliminating the need for extra computations while exhibiting performance enhancements in both grounding masks and referring expressions. With only publicly available training data, our model achieves state-of-the-art results across multiple benchmarks, including diverse modality referring segmentation and region-level referring expression generation., Comment: CVPR 2024

Human mathematicians are often good at recognizing modular and reusable theorems that make complex mathematical results within reach. In this paper, we propose a novel method called theoREm-from-prooF extrACTOR (REFACTOR) for training neural networks to mimic this ability in formal mathematical theorem proving. We show on a set of unseen proofs, REFACTOR is able to extract 19.6% of the theorems that humans would use to write the proofs. When applying the model to the existing Metamath library, REFACTOR extracted 16 new theorems. With newly extracted theorems, we show that the existing proofs in the MetaMath database can be refactored. The new theorems are used very frequently after refactoring, with an average usage of 733.5 times, and help shorten the proof lengths. Lastly, we demonstrate that the prover trained on the new-theorem refactored dataset proves more test theorems and outperforms state-of-the-art baselines by frequently leveraging a diverse set of newly extracted theorems. Code can be found at https://github.com/jinpz/refactor., Comment: ICLR 2024

In the multi-modal domain, the dependence of various models on specific input formats leads to user confusion and hinders progress. To address this challenge, we introduce \textbf{LLMBind}, a novel framework designed to unify a diverse array of multi-modal tasks. By harnessing a Mixture-of-Experts (MoE) Large Language Model (LLM), LLMBind processes multi-modal inputs and generates task-specific tokens, enabling the invocation of corresponding models to accomplish tasks. This unique approach empowers LLMBind to interpret inputs and generate outputs across various modalities, including image, text, video, and audio. Furthermore, we have constructed an interaction dataset comprising 400k instructions, which unlocks the ability of LLMBind for interactive visual generation and editing tasks. Extensive experimentation demonstrates that LLMBind achieves very superior performance across diverse tasks and outperforms existing models in user evaluations conducted in real-world scenarios. Moreover, the adaptability of LLMBind allows for seamless integration with the latest models and extension to new modality tasks, highlighting its potential to serve as a unified AI agent for modeling universal modalities.

Integrated quantum light source is increasingly desirable in large-scale quantum information processing.~Despite recent remarkable advances, new material platform is constantly being explored for the fully on-chip integration of quantum light generation, active and passive manipulation, and detection. Here, for the first time, we demonstrate a gallium nitride (GaN) microring based quantum light generation in the telecom C-band, which has potential towards the monolithic integration of quantum light source.~In our demonstration, the GaN microring has a free spectral range of 330 GHz and a near-zero anomalous dispersion region of over 100 nm. The generation of energy-time entangled photon pair is demonstrated with a typical raw two-photon interference visibility of 95.5$\pm$6.5%, which is further configured to generate heralded single photon with a typical heralded second-order auto-correlation $g^{(2)}_{H}(0)$ of 0.045$\pm$0.001. Our results pave the way for developing chip-scale quantum photonic circuit.

We propose SPHINX-X, an extensive Multimodality Large Language Model (MLLM) series developed upon SPHINX. To improve the architecture and training efficiency, we modify the SPHINX framework by removing redundant visual encoders, bypassing fully-padded sub-images with skip tokens, and simplifying multi-stage training into a one-stage all-in-one paradigm. To fully unleash the potential of MLLMs, we assemble a comprehensive multi-domain and multimodal dataset covering publicly available resources in language, vision, and vision-language tasks. We further enrich this collection with our curated OCR intensive and Set-of-Mark datasets, extending the diversity and generality. By training over different base LLMs including TinyLlama1.1B, InternLM2-7B, LLaMA2-13B, and Mixtral8x7B, we obtain a spectrum of MLLMs that vary in parameter size and multilingual capabilities. Comprehensive benchmarking reveals a strong correlation between the multi-modal performance with the data and parameter scales. Code and models are released at https://github.com/Alpha-VLLM/LLaMA2-Accessory, Comment: Accepted by ICML 2024. Code and models are released at https://github.com/Alpha-VLLM/LLaMA2-Accessory

Machine learning algorithms are increasingly provided as black-box cloud services or pre-trained models, without access to their training data. This motivates the problem of zero-shot out-of-distribution (OOD) detection. Concretely, we aim to detect OOD objects that do not belong to the classifier's label set but are erroneously classified as in-distribution (ID) objects. Our approach, RONIN, uses an off-the-shelf diffusion model to replace detected objects with inpainting. RONIN conditions the inpainting process with the predicted ID label, drawing the input object closer to the in-distribution domain. As a result, the reconstructed object is very close to the original in the ID cases and far in the OOD cases, allowing RONIN to effectively distinguish ID and OOD samples. Throughout extensive experiments, we demonstrate that RONIN achieves competitive results compared to previous approaches across several datasets, both in zero-shot and non-zero-shot settings.

Recent advances demonstrate that scaling Large Vision-Language Models (LVLMs) effectively improves downstream task performances. However, existing scaling methods enable all model parameters to be active for each token in the calculation, which brings massive training and inferring costs. In this work, we propose a simple yet effective training strategy MoE-Tuning for LVLMs. This strategy innovatively addresses the common issue of performance degradation in multi-modal sparsity learning, consequently constructing a sparse model with an outrageous number of parameters but a constant computational cost. Furthermore, we present the MoE-LLaVA, a MoE-based sparse LVLM architecture, which uniquely activates only the top-k experts through routers during deployment, keeping the remaining experts inactive. Extensive experiments show the significant performance of MoE-LLaVA in a variety of visual understanding and object hallucination benchmarks. Remarkably, with only approximately 3B sparsely activated parameters, MoE-LLaVA demonstrates performance comparable to the LLaVA-1.5-7B on various visual understanding datasets and even surpasses the LLaVA-1.5-13B in object hallucination benchmark. Through MoE-LLaVA, we aim to establish a baseline for sparse LVLMs and provide valuable insights for future research in developing more efficient and effective multi-modal learning systems. Code is released at https://github.com/PKU-YuanGroup/MoE-LLaVA., Comment: K = P + N represents the length of the output sequence in the formula (8)

Thermal nonreciprocity plays a vital role in chip heat dissipation, energy-saving design, and high-temperature hyperthermia, typically realized through the use of advanced metamaterials with nonlinear, advective, spatiotemporal, or gradient properties. However, challenges such as fixed structural designs with limited adjustability, high energy consumption, and a narrow operational temperature range remain prevalent. Here, a systematic framework is introduced to achieve reconfigurable, zero-energy, and wide-temperature thermal nonreciprocity by transforming wasteful heat loss into a valuable regulatory tool. Vertical slabs composed of natural bulk materials enable asymmetric heat loss through natural convection, disrupting the inversion symmetry of thermal conduction. The reconfigurability of this system stems from the ability to modify heat loss by adjusting thermal conductivity, size, placement, and quantity of the slabs. Moreover, this structure allows for precise control of zero-energy thermal nonreciprocity across a broad temperature spectrum, utilizing solely environmental temperature gradients without additional energy consumption. This research presents a different approach to achieving nonreciprocity, broadening the potential for nonreciprocal devices such as thermal diodes and topological edge states, and inspiring further exploration of nonreciprocity in other loss-based systems., Comment: 23 pages, 4 figures

The precise measurement of temperature is crucial in various fields such as biology, medicine, industrial automation, energy management, and daily life applications. While in most scenarios, sensors with a fixed thermal conductivity inevitably mismatch the analogous parameter of the medium being measured, thus causing the distortion and inaccurate detection of original temperature fields. Despite recent efforts on addressing the parameter-mismatch issue, all current solutions are constrained to a fixed working medium whereas a more universal sensor should function in a variety of scenes. Here, we report a dynamic thermal sensor capable of highly accurate measurements in diverse working environments. Remarkably, thanks to the highly tunable thermal conductivity of the expanded-plane structure, this sensor works effect on background mediums with a wide range of conductivity. Such a development greatly enhances the robustness and adaptability of thermal sensors, setting a solid foundation for applications in multi-physical sensing scenarios.

Recent one image to 3D generation methods commonly adopt Score Distillation Sampling (SDS). Despite the impressive results, there are multiple deficiencies including multi-view inconsistency, over-saturated and over-smoothed textures, as well as the slow generation speed. To address these deficiencies, we present Repaint123 to alleviate multi-view bias as well as texture degradation and speed up the generation process. The core idea is to combine the powerful image generation capability of the 2D diffusion model and the texture alignment ability of the repainting strategy for generating high-quality multi-view images with consistency. We further propose visibility-aware adaptive repainting strength for overlap regions to enhance the generated image quality in the repainting process. The generated high-quality and multi-view consistent images enable the use of simple Mean Square Error (MSE) loss for fast 3D content generation. We conduct extensive experiments and show that our method has a superior ability to generate high-quality 3D content with multi-view consistency and fine textures in 2 minutes from scratch. Our project page is available at https://pku-yuangroup.github.io/repaint123/., Comment: Project page: https://pku-yuangroup.github.io/repaint123/

Image Retrieval aims to retrieve corresponding images based on a given query. In application scenarios, users intend to express their retrieval intent through various query styles. However, current retrieval tasks predominantly focus on text-query retrieval exploration, leading to limited retrieval query options and potential ambiguity or bias in user intention. In this paper, we propose the Style-Diversified Query-Based Image Retrieval task, which enables retrieval based on various query styles. To facilitate the novel setting, we propose the first Diverse-Style Retrieval dataset, encompassing diverse query styles including text, sketch, low-resolution, and art. We also propose a light-weighted style-diversified retrieval framework. For various query style inputs, we apply the Gram Matrix to extract the query's textural features and cluster them into a style space with style-specific bases. Then we employ the style-init prompt tuning module to enable the visual encoder to comprehend the texture and style information of the query. Experiments demonstrate that our model, employing the style-init prompt tuning strategy, outperforms existing retrieval models on the style-diversified retrieval task. Moreover, style-diversified queries~(sketch+text, art+text, etc) can be simultaneously retrieved in our model. The auxiliary information from other queries enhances the retrieval performance within the respective query., Comment: 16 pages, 7 figures

The Large Vision-Language Model (LVLM) has enhanced the performance of various downstream tasks in visual-language understanding. Most existing approaches encode images and videos into separate feature spaces, which are then fed as inputs to large language models. However, due to the lack of unified tokenization for images and videos, namely misalignment before projection, it becomes challenging for a Large Language Model (LLM) to learn multi-modal interactions from several poor projection layers. In this work, we unify visual representation into the language feature space to advance the foundational LLM towards a unified LVLM. As a result, we establish a simple but robust LVLM baseline, Video-LLaVA, which learns from a mixed dataset of images and videos, mutually enhancing each other. Video-LLaVA achieves superior performances on a broad range of 9 image benchmarks across 5 image question-answering datasets and 4 image benchmark toolkits. Additionally, our Video-LLaVA also outperforms Video-ChatGPT by 5.8%, 9.9%, 18.6%, and 10.1% on MSRVTT, MSVD, TGIF, and ActivityNet, respectively. Notably, extensive experiments demonstrate that Video-LLaVA mutually benefits images and videos within a unified visual representation, outperforming models designed specifically for images or videos. We aim for this work to provide modest insights into the multi-modal inputs for the LLM. Code address: \href{https://github.com/PKU-YuanGroup/Video-LLaVA}

Large language models have demonstrated impressive universal capabilities across a wide range of open-ended tasks and have extended their utility to encompass multimodal conversations. However, existing methods encounter challenges in effectively handling both image and video understanding, particularly with limited visual tokens. In this work, we introduce Chat-UniVi, a Unified Vision-language model capable of comprehending and engaging in conversations involving images and videos through a unified visual representation. Specifically, we employ a set of dynamic visual tokens to uniformly represent images and videos. This representation framework empowers the model to efficiently utilize a limited number of visual tokens to simultaneously capture the spatial details necessary for images and the comprehensive temporal relationship required for videos. Moreover, we leverage a multi-scale representation, enabling the model to perceive both high-level semantic concepts and low-level visual details. Notably, Chat-UniVi is trained on a mixed dataset containing both images and videos, allowing direct application to tasks involving both mediums without requiring any modifications. Extensive experimental results demonstrate that Chat-UniVi consistently outperforms even existing methods exclusively designed for either images or videos. Code is available at https://github.com/PKU-YuanGroup/Chat-UniVi., Comment: Accepted by CVPR 2024 (Highlight)

Most text-driven human motion generation methods employ sequential modeling approaches, e.g., transformer, to extract sentence-level text representations automatically and implicitly for human motion synthesis. However, these compact text representations may overemphasize the action names at the expense of other important properties and lack fine-grained details to guide the synthesis of subtly distinct motion. In this paper, we propose hierarchical semantic graphs for fine-grained control over motion generation. Specifically, we disentangle motion descriptions into hierarchical semantic graphs including three levels of motions, actions, and specifics. Such global-to-local structures facilitate a comprehensive understanding of motion description and fine-grained control of motion generation. Correspondingly, to leverage the coarse-to-fine topology of hierarchical semantic graphs, we decompose the text-to-motion diffusion process into three semantic levels, which correspond to capturing the overall motion, local actions, and action specifics. Extensive experiments on two benchmark human motion datasets, including HumanML3D and KIT, with superior performances, justify the efficacy of our method. More encouragingly, by modifying the edge weights of hierarchical semantic graphs, our method can continuously refine the generated motion, which may have a far-reaching impact on the community. Code and pre-training weights are available at https://github.com/jpthu17/GraphMotion., Comment: Accepted by NeurIPS 2023