Search

The colossal parameters and computational overhead of Large Language Models (LLMs) challenge their real-world applications. Network pruning, which targets unstructured or structured sparsity by removing redundant parameters, has recently been explored for LLM acceleration. Existing LLM pruning works focus on unstructured pruning, which typically requires special hardware support for a practical speed-up. In contrast, structured pruning can reduce latency on general devices. However, it remains a challenge to perform structured pruning efficiently and maintain performance, especially at high sparsity ratios. To this end, we introduce an efficient structured pruning framework named CFSP, which leverages both Coarse (interblock) and Fine-grained (intrablock) activation information as an importance criterion to guide pruning. The pruning is highly efficient, as it only requires one forward pass to compute feature activations. Specifically, we first allocate the sparsity budget across blocks based on their importance and then retain important weights within each block. In addition, we introduce a recovery fine-tuning strategy that adaptively allocates training overhead based on coarse-grained importance to further improve performance. Experimental results demonstrate that CFSP outperforms existing methods on diverse models across various sparsity budgets. Our code will be available at https://github.com/wyxscir/CFSP., Comment: Work in progress

Existing scene text removal (STR) task suffers from insufficient training data due to the expensive pixel-level labeling. In this paper, we aim to address this issue by introducing a Text-aware Masked Image Modeling algorithm (TMIM), which can pretrain STR models with low-cost text detection labels (e.g., text bounding box). Different from previous pretraining methods that use indirect auxiliary tasks only to enhance the implicit feature extraction ability, our TMIM first enables the STR task to be directly trained in a weakly supervised manner, which explores the STR knowledge explicitly and efficiently. In TMIM, first, a Background Modeling stream is built to learn background generation rules by recovering the masked non-text region. Meanwhile, it provides pseudo STR labels on the masked text region. Second, a Text Erasing stream is proposed to learn from the pseudo labels and equip the model with end-to-end STR ability. Benefiting from the two collaborative streams, our STR model can achieve impressive performance only with the public text detection datasets, which greatly alleviates the limitation of the high-cost STR labels. Experiments demonstrate that our method outperforms other pretrain methods and achieves state-of-the-art performance (37.35 PSNR on SCUT-EnsText). Code will be available at https://github.com/wzx99/TMIM., Comment: Accepted by ECCV 2024

Current data compression methods, such as sparsification in Federated Averaging (FedAvg), effectively enhance the communication efficiency of Federated Learning (FL). However, these methods encounter challenges such as the straggler problem and diminished model performance due to heterogeneous bandwidth and non-IID (Independently and Identically Distributed) data. To address these issues, we introduce a bandwidth-aware compression framework for FL, aimed at improving communication efficiency while mitigating the problems associated with non-IID data. First, our strategy dynamically adjusts compression ratios according to bandwidth, enabling clients to upload their models at a close pace, thus exploiting the otherwise wasted time to transmit more data. Second, we identify the non-overlapped pattern of retained parameters after compression, which results in diminished client update signals due to uniformly averaged weights. Based on this finding, we propose a parameter mask to adjust the client-averaging coefficients at the parameter level, thereby more closely approximating the original updates, and improving the training convergence under heterogeneous environments. Our evaluations reveal that our method significantly boosts model accuracy, with a maximum improvement of 13% over the uncompressed FedAvg. Moreover, it achieves a $3.37\times$ speedup in reaching the target accuracy compared to FedAvg with a Top-K compressor, demonstrating its effectiveness in accelerating convergence with compression. The integration of common compression techniques into our framework further establishes its potential as a versatile foundation for future cross-device, communication-efficient FL research, addressing critical challenges in FL and advancing the field of distributed machine learning.

Data serves as the fundamental foundation for advancing deep learning, particularly tabular data presented in a structured format, which is highly conducive to modeling. However, even in the era of LLM, obtaining tabular data from sensitive domains remains a challenge due to privacy or copyright concerns. Hence, exploring how to effectively use models like LLMs to generate realistic and privacy-preserving synthetic tabular data is urgent. In this paper, we take a step forward to explore LLMs for tabular data synthesis and privacy protection, by introducing a new framework HARMONIC for tabular data generation and evaluation. In the tabular data generation of our framework, unlike previous small-scale LLM-based methods that rely on continued pre-training, we explore the larger-scale LLMs with fine-tuning to generate tabular data and enhance privacy. Based on idea of the k-nearest neighbors algorithm, an instruction fine-tuning dataset is constructed to inspire LLMs to discover inter-row relationships. Then, with fine-tuning, LLMs are trained to remember the format and connections of the data rather than the data itself, which reduces the risk of privacy leakage. In the evaluation part of our framework, we develop specific privacy risk metrics DLT for LLM synthetic data generation, as well as performance evaluation metrics LLE for downstream LLM tasks. Our experiments find that this tabular data generation framework achieves equivalent performance to existing methods with better privacy, which also demonstrates our evaluation framework for the effectiveness of synthetic data and privacy risks in LLM scenarios.

The Ruddlesden-Popper (R-P) bilayer nickelate, La3Ni2O7, was recently found to show signatures of high-temperature superconductivity (HTSC) at pressures above 14 GPa. Subsequent investigations achieved zero resistance in single- and poly-crystalline samples under hydrostatic pressure conditions. Yet, obvious diamagnetic signals, the other hallmark of superconductors, are still lacking owing to the filamentary nature with low superconducting volume fraction. The presence of a novel "1313" polymorph and competing R-P phases obscured proper identification of the phase for HTSC. Thus, achieving bulk HTSC and identifying the phase at play are the most prominent tasks at present. Here, we address these issues in the praseodymium (Pr)-doped La2PrNi2O7 polycrystalline samples. We find that the substitutions of Pr for La effectively inhibits the intergrowth of different R-P phases, resulting in nearly pure bilayer structure. For La2PrNi2O7, pressure-induced orthorhombic-to-tetragonal structural transition takes place at Pc ~ 11 GPa, above which HTSC emerges gradually upon further compression. The superconducting transition temperatures at 18-20 GPa reach Tconset = 82.5 K and Tczero = 60 K, which are the highest values among known nickelate superconductors. More importantly, bulk HTSC was testified by detecting clear diamagnetic signals below ~75 K corresponding to an estimated superconducting volume fraction ~ 57(5)% at 20 GPa. Our results not only resolve the existing controversies but also illuminate directions for exploring bulk HTSC in the bilayer nickelates.

We introduce the notions of de Bruijn polyominoes and prismatic polyominoes, which generalize the notions of de Bruijn sequences and arrays. Given a small fixed polyomino $p$ and a set of colors $[n]$, a de Bruijn polyomino for $(p,n)$ is a colored fixed polyomino $P$ with cells colored from $[n]$ such that every possible coloring of $p$ from $[n]$ exists as a subset of $P$. We call de Bruijn polyominoes for $(p,n)$ of minimum size $(p,n)$-prismatic. We discuss for some values of $p$ and $n$ the shape of a $(p,n)$-prismatic polyomino $P$, the construction of a coloring of $P$, and the enumeration of the colorings of $P$. We find evidence that the difficulty of these problems may depend on the parity of the size of $p$

Mental manipulation, a significant form of abuse in interpersonal conversations, presents a challenge to identify due to its context-dependent and often subtle nature. The detection of manipulative language is essential for protecting potential victims, yet the field of Natural Language Processing (NLP) currently faces a scarcity of resources and research on this topic. Our study addresses this gap by introducing a new dataset, named ${\rm M{\small ental}M{\small anip}}$, which consists of $4,000$ annotated movie dialogues. This dataset enables a comprehensive analysis of mental manipulation, pinpointing both the techniques utilized for manipulation and the vulnerabilities targeted in victims. Our research further explores the effectiveness of leading-edge models in recognizing manipulative dialogue and its components through a series of experiments with various configurations. The results demonstrate that these models inadequately identify and categorize manipulative content. Attempts to improve their performance by fine-tuning with existing datasets on mental health and toxicity have not overcome these limitations. We anticipate that ${\rm M{\small ental}M{\small anip}}$ will stimulate further research, leading to progress in both understanding and mitigating the impact of mental manipulation in conversations., Comment: Accepted at ACL 2024

We present a novel approach to predicting the pressure and flow rate of flexible electrohydrodynamic pumps using the Kolmogorov-Arnold Network. Inspired by the Kolmogorov-Arnold representation theorem, KAN replaces fixed activation functions with learnable spline-based activation functions, enabling it to approximate complex nonlinear functions more effectively than traditional models like Multi-Layer Perceptron and Random Forest. We evaluated KAN on a dataset of flexible EHD pump parameters and compared its performance against RF, and MLP models. KAN achieved superior predictive accuracy, with Mean Squared Errors of 12.186 and 0.001 for pressure and flow rate predictions, respectively. The symbolic formulas extracted from KAN provided insights into the nonlinear relationships between input parameters and pump performance. These findings demonstrate that KAN offers exceptional accuracy and interpretability, making it a promising alternative for predictive modeling in electrohydrodynamic pumping.

In text recognition, self-supervised pre-training emerges as a good solution to reduce dependence on expansive annotated real data. Previous studies primarily focus on local visual representation by leveraging mask image modeling or sequence contrastive learning. However, they omit modeling the linguistic information in text images, which is crucial for recognizing text. To simultaneously capture local character features and linguistic information in visual space, we propose Symmetric Superimposition Modeling (SSM). The objective of SSM is to reconstruct the direction-specific pixel and feature signals from the symmetrically superimposed input. Specifically, we add the original image with its inverted views to create the symmetrically superimposed inputs. At the pixel level, we reconstruct the original and inverted images to capture character shapes and texture-level linguistic context. At the feature level, we reconstruct the feature of the same original image and inverted image with different augmentations to model the semantic-level linguistic context and the local character discrimination. In our design, we disrupt the character shape and linguistic rules. Consequently, the dual-level reconstruction facilitates understanding character shapes and linguistic information from the perspective of visual texture and feature semantics. Experiments on various text recognition benchmarks demonstrate the effectiveness and generality of SSM, with 4.1% average performance gains and 86.6% new state-of-the-art average word accuracy on Union14M benchmarks. The code is available at https://github.com/FaltingsA/SSM., Comment: Accepted to IJCAI2024

Scene text images contain not only style information (font, background) but also content information (character, texture). Different scene text tasks need different information, but previous representation learning methods use tightly coupled features for all tasks, resulting in sub-optimal performance. We propose a Disentangled Representation Learning framework (DARLING) aimed at disentangling these two types of features for improved adaptability in better addressing various downstream tasks (choose what you really need). Specifically, we synthesize a dataset of image pairs with identical style but different content. Based on the dataset, we decouple the two types of features by the supervision design. Clearly, we directly split the visual representation into style and content features, the content features are supervised by a text recognition loss, while an alignment loss aligns the style features in the image pairs. Then, style features are employed in reconstructing the counterpart image via an image decoder with a prompt that indicates the counterpart's content. Such an operation effectively decouples the features based on their distinctive properties. To the best of our knowledge, this is the first time in the field of scene text that disentangles the inherent properties of the text images. Our method achieves state-of-the-art performance in Scene Text Recognition, Removal, and Editing., Comment: Accepted to CVPR 2024

We introduce a novel time-division multiplexing differential saturated absorption spectroscopy (TDMDSAS) approach, providing superior accuracy and stability in Doppler-free spectroscopy. By distinguishing probe and reference fields in the temporal domain, TDMDSAS efficiently suppresses Doppler broadening and common-mode optical noise. We utilized this technology to determine the absolute frequency of diverse neutral Yb isotopes across its $6s^2\ ^{1}S_0\to 6s6p ^{1}P_1$ transitions. Furthermore, the first-ever observation of in-situ Doppler-free Zeeman sub-level spectra was accomplished, enabling the determination of magnetic field gradients. We stabilized a UV diode laser at 399 nm using an error signal derived from the spectral first-derivative demodulated signal of $^{174}\mathrm{Yb}$. This technique yielded a frequency stability of up to 15 kHz with a 40 s averaging time and a standard deviation of around 180 kHz over a half-hour period. Given its low cost, straightforward, and scalable nature, TDMDSAS holds excellent potential in metrology and quantum applications.

This paper tackles the intricate challenge of object removal to update the radiance field using the 3D Gaussian Splatting. The main challenges of this task lie in the preservation of geometric consistency and the maintenance of texture coherence in the presence of the substantial discrete nature of Gaussian primitives. We introduce a robust framework specifically designed to overcome these obstacles. The key insight of our approach is the enhancement of information exchange among visible and invisible areas, facilitating content restoration in terms of both geometry and texture. Our methodology begins with optimizing the positioning of Gaussian primitives to improve geometric consistency across both removed and visible areas, guided by an online registration process informed by monocular depth estimation. Following this, we employ a novel feature propagation mechanism to bolster texture coherence, leveraging a cross-attention design that bridges sampling Gaussians from both uncertain and certain areas. This innovative approach significantly refines the texture coherence within the final radiance field. Extensive experiments validate that our method not only elevates the quality of novel view synthesis for scenes undergoing object removal but also showcases notable efficiency gains in training and rendering speeds., Comment: Project Page: https://w-ted.github.io/publications/gscream

The swift evolution of Large-scale Models (LMs), either language-focused or multi-modal, has garnered extensive attention in both academy and industry. But despite the surge in interest in this rapidly evolving area, there are scarce systematic reviews on their capabilities and potential in distinct impactful scenarios. This paper endeavours to help bridge this gap, offering a thorough examination of the current landscape of LM usage in regards to complex game playing scenarios and the challenges still open. Here, we seek to systematically review the existing architectures of LM-based Agents (LMAs) for games and summarize their commonalities, challenges, and any other insights. Furthermore, we present our perspective on promising future research avenues for the advancement of LMs in games. We hope to assist researchers in gaining a clear understanding of the field and to generate more interest in this highly impactful research direction. A corresponding resource, continuously updated, can be found in our GitHub repository., Comment: 13 pages, 3 figures

Recently, MLP structures have regained popularity, with MLP-Mixer standing out as a prominent example. In the field of computer vision, MLP-Mixer is noted for its ability to extract data information from both channel and token perspectives, effectively acting as a fusion of channel and token information. Indeed, Mixer represents a paradigm for information extraction that amalgamates channel and token information. The essence of Mixer lies in its ability to blend information from diverse perspectives, epitomizing the true concept of "mixing" in the realm of neural network architectures. Beyond channel and token considerations, it is possible to create more tailored mixers from various perspectives to better suit specific task requirements. This study focuses on the domain of audio recognition, introducing a novel model named Audio Spectrogram Mixer with Roll-Time and Hermit FFT (ASM-RH) that incorporates insights from both time and frequency domains. Experimental results demonstrate that ASM-RH is particularly well-suited for audio data and yields promising outcomes across multiple classification tasks. The models and optimal weights files will be published.

As a promising 6G technology, integrated sensing and communication (ISAC) gains growing interest. ISAC provides integration gain via sharing spectrum, hardware, and software. However, concerns exist regarding its sensing performance when compared to dedicated radar systems. To address this issue, the advantages of widely deployed networks should be utilized, and this paper proposes networked collaborative sensing (NCS) using multi-domain measurements (MM), including range, Doppler, and two-dimension angle of arrival. In the NCS-MM architecture, this paper proposes a novel multi-domain decoupling model and a novel guard band-based protocol. The proposed model simplifies multi-domain derivations and algorithm designs, and the proposed protocol conserves resources and mitigates NCS interference. To determine the performance limits, this paper derives the Cram\'er-Rao lower bound (CRLB) of three-dimension position and velocity in NCS-MM. An accumulated single-dimension channel model is used to obtain the CRLB of MM, which is proven to be equivalent to that of the multi-dimension model. The algorithms of both MM estimation and fusion are proposed. An arbitrary-dimension Newtonized orthogonal matched pursuit (AD-NOMP) is proposed to accurately estimate grid-less MM. The degree-of-freedom (DoF) of MM is analyzed, and a novel DoF-based two-stage weighted least squares (TSWLS) is proposed to reduce equations without DoF loss. The numerical results show that the performances of the proposed algorithms are close to their performance limits.

Due to the inability to receive signals from the Global Navigation Satellite System (GNSS) in extreme conditions, achieving accurate and robust navigation for Unmanned Aerial Vehicles (UAVs) is a challenging task. Recently emerged, vision-based navigation has been a promising and feasible alternative to GNSS-based navigation. However, existing vision-based techniques are inadequate in addressing flight deviation caused by environmental disturbances and inaccurate position predictions in practical settings. In this paper, we present a novel angle robustness navigation paradigm to deal with flight deviation in point-to-point navigation tasks. Additionally, we propose a model that includes the Adaptive Feature Enhance Module, Cross-knowledge Attention-guided Module and Robust Task-oriented Head Module to accurately predict direction angles for high-precision navigation. To evaluate the vision-based navigation methods, we collect a new dataset termed as UAV_AR368. Furthermore, we design the Simulation Flight Testing Instrument (SFTI) using Google Earth to simulate different flight environments, thereby reducing the expenses associated with real flight testing. Experiment results demonstrate that the proposed model outperforms the state-of-the-art by achieving improvements of 26.0% and 45.6% in the success rate of arrival under ideal and disturbed circumstances, respectively., Comment: 9 pages, 4 figures

Serving systems for Large Language Models (LLMs) are often optimized to improve quality of service (QoS) and throughput. However, due to the lack of open-sourced LLM serving workloads, these systems are frequently evaluated under unrealistic workload assumptions. Consequently, performance may degrade when these systems are deployed in real-world scenarios. This work presents BurstGPT, an LLM serving workload with 5.29 million traces from regional Azure OpenAI GPT services over 121 days. BurstGPT captures realistic LLM serving characteristics through detailed tracing of: (1) Concurrency of requests: It traces burstiness variations of requests in Azure OpenAI GPT services, revealing diversified concurrency patterns in different services and model types. (2) Response Lengths of requests: It traces the auto-regressive serving processes of GPT models, showing statistical relations between requests and their responses. (3) Failures of requests: It traces failures of conversation and API services, showing intensive resource needs and limited resource availability of such services in Azure. Details of the characteristics can serve multiple purposes in LLM serving optimizations, such as system evaluation and trace provisioning. In our demo evaluation with BurstGPT, we observe that frequent variations in BurstGPT reveal declines in efficiency, stability, or reliability in realistic LLM serving. We identify that the generalization of KV cache management and request scheduling optimization is not guaranteed for different workloads, especially when systems are poorly optimized for unrealistic workloads. We have made the dataset publicly available to encourage further research at https://github.com/HPMLL/BurstGPT.

We carry out a systematic study of the electronic and magnetic structure of the ambient-pressure bilayer La$_3$Ni$_2$O$_7$. Employing the hybrid exchange-correlation functional, we show that the exchange-correlation pushes the antibonding $d_{z^2}$ bands below the Fermi level to be fully occupied. The calculated Fermi surfaces and the correlation normalized band structure match well with the experimental findings at ambient pressure. Moreover, the electronic susceptibility calculated for this new band structure features nesting-induced peaks near the wave vector $Q=(\pi/2, \pi/2)$, suggesting a possible density wave instability in agreement with recent experiments. Through a mean field study and DFT+U calculation, we confirm the spin-charge intertwined double stripe order is the magnetic ground state. Our results provide a faithful description for the low-pressure La$_3$Ni$_2$O$_7$ electronic structure., Comment: 4 pages, 7 figures and supplemental material

Transformer structures have demonstrated outstanding skills in the deep learning space recently, significantly increasing the accuracy of models across a variety of domains. Researchers have started to question whether such a sophisticated network structure is actually necessary and whether equally outstanding results can be reached with reduced inference cost due to its complicated network topology and high inference cost. In order to prove the Mixer's efficacy on three datasets Speech Commands, UrbanSound8k, and CASIA Chinese Sentiment Corpus this paper applies amore condensed version of the Mixer to an audio classification task and conducts comparative experiments with the Transformer-based Audio Spectrogram Transformer (AST)model. In addition, this paper conducts comparative experiments on the application of several activation functions in Mixer, namely GeLU, Mish, Swish and Acon-C. Further-more, the use of various activation functions in Mixer, including GeLU, Mish, Swish, and Acon-C, is compared in this research through comparison experiments. Additionally, some AST model flaws are highlighted, and the model suggested in this study is improved as a result. In conclusion, a model called the Audio Spectrogram Mixer, which is the first model for audio classification with Mixer, is suggested in this study and the model's future directions for improvement are examined.

Transformer requires a fixed number of layers and heads which makes them inflexible to the complexity of individual samples and expensive in training and inference. To address this, we propose a sample-based Dynamic Hierarchical Transformer (DHT) model whose layers and heads can be dynamically configured with single data samples via solving contextual bandit problems. To determine the number of layers and heads, we use the Uniform Confidence Bound while we deploy combinatorial Thompson Sampling in order to select specific head combinations given their number. Different from previous work that focuses on compressing trained networks for inference only, DHT is not only advantageous for adaptively optimizing the underlying network architecture during training but also has a flexible network for efficient inference. To the best of our knowledge, this is the first comprehensive data-driven dynamic transformer without any additional auxiliary neural networks that implement the dynamic system. According to the experiment results, we achieve up to 74% computational savings for both training and inference with a minimal loss of accuracy., Comment: We miss some authorship information. And miss some important information in references

Due to the necessary security for the airport and flight, passengers are required to have strict security check before getting aboard. However, there are frequent complaints of wasting huge amount of time while waiting for the security check. This paper presents a potential solution aimed at optimizing gate setup procedures specifically tailored for Chicago OHare International Airport. By referring to queueing theory and performing Monte Carlo simulations, we propose an approach to significantly diminish the average waiting time to a more manageable level. Additionally, our study meticulously examines and identifies the influential factors contributing to this optimization, providing a comprehensive understanding of their impact.

Federated learning (FL) is an emerging paradigm for decentralized training of machine learning models on distributed clients, without revealing the data to the central server. The learning scheme may be horizontal, vertical or hybrid (both vertical and horizontal). Most existing research work with deep neural network (DNN) modelling is focused on horizontal data distributions, while vertical and hybrid schemes are much less studied. In this paper, we propose a generalized algorithm FedEmb, for modelling vertical and hybrid DNN-based learning. The idea of our algorithm is characterised by higher inference accuracy, stronger privacy-preserving properties, and lower client-server communication bandwidth demands as compared with existing work. The experimental results show that FedEmb is an effective method to tackle both split feature & subject space decentralized problems, shows 0.3% to 4.2% inference accuracy improvement with limited privacy revealing for datasets stored in local clients, and reduces 88.9 % time complexity over vertical baseline method., Comment: Miss some important information and references. The publication hasn't been online in the journal

Spectrum sensing technology is a crucial aspect of modern communication technology, serving as one of the essential techniques for efficiently utilizing scarce information resources in tight frequency bands. This paper first introduces three common logical circuit decision criteria in hard decisions and analyzes their decision rigor. Building upon hard decisions, the paper further introduces a method for multi-user spectrum sensing based on soft decisions. Then the paper simulates the false alarm probability and detection probability curves corresponding to the three criteria. The simulated results of multi-user collaborative sensing indicate that the simulation process significantly reduces false alarm probability and enhances detection probability. This approach effectively detects spectrum resources unoccupied during idle periods, leveraging the concept of time-division multiplexing and rationalizing the redistribution of information resources. The entire computation process relies on the calculation principles of power spectral density in communication theory, involving threshold decision detection for noise power and the sum of noise and signal power. It provides a secondary decision detection, reflecting the perceptual decision performance of logical detection methods with relative accuracy., Comment: https://aei.ewapublishing.org/article.html?pk=e24c40d220434209ae2fe2e984bcf2c2

The recent discovery of high-temperature superconductivity in the Ruddlesden-Popper phase La3Ni2O7 under high pressure marks a significant breakthrough in the field of 3d transition-metal oxide superconductors. For an emerging novel class of high-Tc superconductors, it is crucial to find more analogous superconducting materials with a dedicated effort toward broadening the scope of nickelate superconductors. Here, we report on the observation of high-Tc superconductivity in the high-pressure tetragonal I4/mmm phase of La2PrNi2O7 above ~10 GPa, which is distinct from the reported orthorhombic Fmmm phase of La3Ni2O7 above 14 GPa. For La2PrNi2O7, the onset and the zero-resistance temperatures of superconductivity reach Tconset = 78.2 K and Tczero = 40 K at 15 GPa. This superconducting phase shares the samilar structural symmetry as many cuprate superconductors, providing a fresh platform to investigate underlying mechanisms of nickelate superconductors., Comment: 19 pages and 6 figures

To efficiently scale large model (LM) training, researchers transition from data parallelism (DP) to hybrid parallelism (HP) on GPU clusters, which frequently experience hardware and software failures. Existing works introduce in-memory checkpointing optimizations that snapshot parameters to device memory for rapid failure recovery. However, these methods introduce severe resource competition between checkpointing and training, which can work under DP but can hardly scale under resource-intensive HP. To ensure low checkpointing overhead for hybrid-parallel training, this paper introduces a distributed in-memory checkpointing system with near-zero in-memory saving overhead. It strives from two aspects to mitigate the on-host resource competition caused by in-memory checkpointing: (1) It introduces Hierarchical Asynchronous Snapshotting Coordination in the checkpoint saving stage. This approach uses three-level asynchronous on-device scheduling to enhance parallelism between snapshotting and training, thereby minimizing snapshotting overhead. (2) It proposes Hybrid In-memory Checkpoint Protection to enhance checkpoint completeness during hardware failures. Unlike methods that require inter-node communications, which may block training under HP, it creates intra-node redundancy with efficient resource utilization, protecting training against hardware failures with minimal overhead. With these methods, this work enables fast restart for failed HP training with Distributed In-memory Checkpoint Loading, bypassing inefficiencies in NFS reads. In our evaluation, we achieve zero in-memory checkpoint saving overhead on Frontier while training Llama-2-34B on 256 MI250X devices (512 GPUs)., Comment: Fault Tolerance, Checkpoint Optimization, Large Language Model, Foundation Model, Hybrid parallelism

Graph neural networks (GNNs) for link prediction can loosely be divided into two broad categories. First, \emph{node-wise} architectures pre-compute individual embeddings for each node that are later combined by a simple decoder to make predictions. While extremely efficient at inference time (since node embeddings are only computed once and repeatedly reused), model expressiveness is limited such that isomorphic nodes contributing to candidate edges may not be distinguishable, compromising accuracy. In contrast, \emph{edge-wise} methods rely on the formation of edge-specific subgraph embeddings to enrich the representation of pair-wise relationships, disambiguating isomorphic nodes to improve accuracy, but with the cost of increased model complexity. To better navigate this trade-off, we propose a novel GNN architecture whereby the \emph{forward pass} explicitly depends on \emph{both} positive (as is typical) and negative (unique to our approach) edges to inform more flexible, yet still cheap node-wise embeddings. This is achieved by recasting the embeddings themselves as minimizers of a forward-pass-specific energy function (distinct from the actual training loss) that favors separation of positive and negative samples. As demonstrated by extensive empirical evaluations, the resulting architecture retains the inference speed of node-wise models, while producing competitive accuracy with edge-wise alternatives., Comment: 19 pages, 5 figures

In this paper, we explore the potential of the Contrastive Language-Image Pretraining (CLIP) model in scene text recognition (STR), and establish a novel Symmetrical Linguistic Feature Distillation framework (named CLIP-OCR) to leverage both visual and linguistic knowledge in CLIP. Different from previous CLIP-based methods mainly considering feature generalization on visual encoding, we propose a symmetrical distillation strategy (SDS) that further captures the linguistic knowledge in the CLIP text encoder. By cascading the CLIP image encoder with the reversed CLIP text encoder, a symmetrical structure is built with an image-to-text feature flow that covers not only visual but also linguistic information for distillation.Benefiting from the natural alignment in CLIP, such guidance flow provides a progressive optimization objective from vision to language, which can supervise the STR feature forwarding process layer-by-layer.Besides, a new Linguistic Consistency Loss (LCL) is proposed to enhance the linguistic capability by considering second-order statistics during the optimization. Overall, CLIP-OCR is the first to design a smooth transition between image and text for the STR task.Extensive experiments demonstrate the effectiveness of CLIP-OCR with 93.8% average accuracy on six popular STR benchmarks.Code will be available at https://github.com/wzx99/CLIPOCR., Comment: Accepted by ACM MM 2023

Integrated sensing and communication (ISAC) is a key enabler of 6G. Unlike communication radio links, the sensing signal requires to experience round trips from many scatters. Therefore, sensing is more power-sensitive and faces a severer multi-target interference. In this paper, the ISAC system employs dedicated sensing signals, which can be reused as the communication reference signal. This paper proposes to add time-frequency matched windows at both the transmitting and receiving sides, which avoids mismatch loss and increases energy efficiency. Discrete non-linear frequency modulation (DNLFM) is further proposed to achieve both time-domain constant modulus and frequency-domain arbitrary windowing weights. DNLFM uses very few Newton iterations and a simple geometrically-equivalent method to generate, which greatly reduces the complex numerical integral in the conventional method. Moreover, the spatial-domain matched window is proposed to achieve low sidelobes. The simulation results show that the proposed methods gain a higher energy efficiency than conventional methods.

The rapid growth of memory and computation requirements of large language models (LLMs) has outpaced the development of hardware, hindering people who lack large-scale high-end GPUs from training or deploying LLMs. However, consumer-level GPUs, which constitute a larger market share, are typically overlooked in LLM due to their weaker computing performance, smaller storage capacity, and lower communication bandwidth. Additionally, users may have privacy concerns when interacting with remote LLMs. In this paper, we envision a decentralized system unlocking the potential vast untapped consumer-level GPUs in pre-training, inference and fine-tuning of LLMs with privacy protection. However, this system faces critical challenges, including limited CPU and GPU memory, low network bandwidth, the variability of peer and device heterogeneity. To address these challenges, our system design incorporates: 1) a broker with backup pool to implement dynamic join and quit of computing providers; 2) task scheduling with hardware performance to improve system efficiency; 3) abstracting ML procedures into directed acyclic graphs (DAGs) to achieve model and task universality; 4) abstracting intermediate represention and execution planes to ensure compatibility of various devices and deep learning (DL) frameworks. Our performance analysis demonstrates that 50 RTX 3080 GPUs can achieve throughputs comparable to those of 4 H100 GPUs, which are significantly more expensive.

The Noisy Max mechanism and its variations are fundamental private selection algorithms that are used to select items from a set of candidates (such as the most common diseases in a population), while controlling the privacy leakage in the underlying data. A recently proposed extension, Noisy Top-k with Gap, provides numerical information about how much better the selected items are compared to the non-selected items (e.g., how much more common are the selected diseases). This extra information comes at no privacy cost but crucially relies on infinite precision for the privacy guarantees. In this paper, we provide a finite-precision secure implementation of this algorithm that takes advantage of integer arithmetic., Comment: 21 pages

Implicit neural representations have shown powerful capacity in modeling real-world 3D scenes, offering superior performance in novel view synthesis. In this paper, we target a more challenging scenario, i.e., joint scene novel view synthesis and editing based on implicit neural scene representations. State-of-the-art methods in this direction typically consider building separate networks for these two tasks (i.e., view synthesis and editing). Thus, the modeling of interactions and correlations between these two tasks is very limited, which, however, is critical for learning high-quality scene representations. To tackle this problem, in this paper, we propose a unified Neural Radiance Field (NeRF) framework to effectively perform joint scene decomposition and composition for modeling real-world scenes. The decomposition aims at learning disentangled 3D representations of different objects and the background, allowing for scene editing, while scene composition models an entire scene representation for novel view synthesis. Specifically, with a two-stage NeRF framework, we learn a coarse stage for predicting a global radiance field as guidance for point sampling, and in the second fine-grained stage, we perform scene decomposition by a novel one-hot object radiance field regularization module and a pseudo supervision via inpainting to handle ambiguous background regions occluded by objects. The decomposed object-level radiance fields are further composed by using activations from the decomposition module. Extensive quantitative and qualitative results show the effectiveness of our method for scene decomposition and composition, outperforming state-of-the-art methods for both novel-view synthesis and editing tasks., Comment: ICCV2023, Project Page: https://w-ted.github.io/publications/udc-nerf