arXiv subjects

Yutong He

Publications and source records attributed to Yutong He.

At least 37 records · Page 2Linked to original sources

arXiv2025

An All-Reduce Compatible Top-K Compressor for Communication-Efficient Distributed Learning

Communication remains a central bottleneck in large-scale distributed machine learning, and gradient sparsification has emerged as a promising strategy to alleviate this challenge. However, existing gradient compressors face notable limitations: Rand-$K$ discards structural information and performs poorly in practice, while Top-$K$ preserves informative entries but loses the contraction property and requires costly All-Gather operations. In this paper, we propose ARC-Top-$K$, an {All-Reduce}-Compatible Top-$K$ compressor that aligns sparsity patterns across nodes using a lightweight sketch of the gradient, enabling index-free All-Reduce while preserving globally significant information. ARC-Top-$K$ is provably contractive and, when combined with momentum error feedback (EF21M), achieves linear speedup and sharper convergence rates than the original EF21M under standard assumptions. Empirically, ARC-Top-$K$ matches the accuracy of Top-$K$ while reducing wall-clock training time by up to 60.7\%, offering an efficient and scalable solution that combines the robustness of Rand-$K$ with the strong performance of Top-$K$.

cs.LG↗

arXiv2025

Greedy Low-Rank Gradient Compression for Distributed Learning with Convergence Guarantees

Distributed optimization is pivotal for large-scale signal processing and machine learning, yet communication overhead remains a major bottleneck. Low-rank gradient compression, in which the transmitted gradients are approximated by low-rank matrices to reduce communication, offers a promising remedy. Existing methods typically adopt either randomized or greedy compression strategies: randomized approaches project gradients onto randomly chosen subspaces, introducing high variance and degrading empirical performance; greedy methods select the most informative subspaces, achieving strong empirical results but lacking convergence guarantees. To address this gap, we propose GreedyLore--the first Greedy Low-Rank gradient compression algorithm for distributed learning with rigorous convergence guarantees. GreedyLore incorporates error feedback to correct the bias introduced by greedy compression and introduces a semi-lazy subspace update that ensures the compression operator remains contractive throughout all iterations. With these techniques, we prove that GreedyLore achieves a convergence rate of $\mathcal{O}(σ/\sqrt{NT} + 1/T)$ under standard optimizers such as MSGD and Adam--marking the first linear speedup convergence rate for low-rank gradient compression. Extensive experiments are conducted to validate our theoretical findings.

cs.LG↗

arXiv2025

MeCeFO: Enhancing LLM Training Robustness via Fault-Tolerant Optimization

As distributed optimization scales to meet the demands of Large Language Model (LLM) training, hardware failures become increasingly non-negligible. Existing fault-tolerant training methods often introduce significant computational or memory overhead, demanding additional resources. To address this challenge, we propose Memory- and Computation-efficient Fault-tolerant Optimization (MeCeFO), a novel algorithm that ensures robust training with minimal overhead. When a computing node fails, MeCeFO seamlessly transfers its training task to a neighboring node while employing memory- and computation-efficient algorithmic optimizations to minimize the extra workload imposed on the neighboring node handling both tasks. MeCeFO leverages three key algorithmic designs: (i) Skip-connection, which drops the multi-head attention (MHA) module during backpropagation for memory- and computation-efficient approximation; (ii) Recomputation, which reduces activation memory in feedforward networks (FFNs); and (iii) Low-rank gradient approximation, enabling efficient estimation of FFN weight matrix gradients. Theoretically, MeCeFO matches the convergence rate of conventional distributed training, with a rate of $\mathcal{O}(1/\sqrt{nT})$, where n is the data parallelism size and T is the number of iterations. Empirically, MeCeFO maintains robust performance under high failure rates, incurring only a 4.18% drop in throughput, demonstrating 5.0$\times$ to 6.7$\times$ greater resilience than previous SOTA approaches. Codes are available at https://github.com/pkumelon/MeCeFO.

cs.DC↗

arXiv2025

Automated Black-box Prompt Engineering for Personalized Text-to-Image Generation

Prompt engineering is an effective but labor-intensive way to control text-to-image (T2I) generative models. Its time-intensive nature and complexity have spurred the development of algorithms for automated prompt generation. However, these methods often struggle with transferability across T2I models, require white-box access to the underlying model, or produce non-intuitive prompts. In this work, we introduce PRISM, an algorithm that automatically produces human-interpretable and transferable prompts that can effectively generate desired concepts given only black-box access to T2I models. Inspired by large language model (LLM) jailbreaking, PRISM leverages the in-context learning ability of LLMs to iteratively refine the candidate prompt distribution built upon the reference images. Our experiments demonstrate the versatility and effectiveness of PRISM in generating accurate prompts for objects, styles, and images across multiple T2I models, including Stable Diffusion, DALL-E, and Midjourney.

cs.CV↗

arXiv2025

Subspace Optimization for Large Language Models with Convergence Guarantees

Subspace optimization algorithms, such as GaLore (Zhao et al., 2024), have gained attention for pre-training and fine-tuning large language models (LLMs) due to their memory efficiency. However, their convergence guarantees remain unclear, particularly in stochastic settings. In this paper, we reveal that GaLore does not always converge to the optimal solution and provide an explicit counterexample to support this finding. We further explore the conditions under which GaLore achieves convergence, showing that it does so when either (i) a sufficiently large mini-batch size is used or (ii) the gradient noise is isotropic. More significantly, we introduce GoLore (Gradient random Low-rank projection), a novel variant of GaLore that provably converges in typical stochastic settings, even with standard batch sizes. Our convergence analysis extends naturally to other subspace optimization algorithms. Finally, we empirically validate our theoretical results and thoroughly test the proposed mechanisms. Codes are available at https://github.com/pkumelon/Golore.

cs.LG↗

arXiv2025

Counterpart-Induced Millisecond-Scale Truncation Mechanism of Fast Radio Bursts

The observed millisecond-scale duration is an essential yet mysterious feature of fast radio bursts (FRBs). In this Letter, we link the observed soft gamma-ray counterpart of FRB 200428 to electron-positron pair cascades driven by Compton scattering and the Breit-Wheeler process. We demonstrate that such pair cascades can truncate FRBs to durations down to millisecond-scale, thereby establishing millisecond-scale upper bounds on their durations. The physical processes involved in the truncation mechanism occur during the propagation of FRBs after their production. Consequently, this mechanism is independent of the specific production mechanism or origin of the FRBs, suggesting that it could potentially operate in all FRBs. Our results lift the constraint on FRB production mechanisms that they must inherently generate bursts lasting only milliseconds.

astro-ph.HE↗

arXiv2025

Prompt Recovery for Image Generation Models: A Comparative Study of Discrete Optimizers

Recovering natural language prompts for image generation models, solely based on the generated images is a difficult discrete optimization problem. In this work, we present the first head-to-head comparison of recent discrete optimization techniques for the problem of prompt inversion. We evaluate Greedy Coordinate Gradients (GCG), PEZ , Random Search, AutoDAN and BLIP2's image captioner across various evaluation metrics related to the quality of inverted prompts and the quality of the images generated by the inverted prompts. We find that focusing on the CLIP similarity between the inverted prompts and the ground truth image acts as a poor proxy for the similarity between ground truth image and the image generated by the inverted prompts. While the discrete optimizers effectively minimize their objectives, simply using responses from a well-trained captioner often leads to generated images that more closely resemble those produced by the original prompts.

cs.CV↗

arXiv2025

Performance-Aware Control of Modular Batteries For Fast Frequency Response

Modular batteries can be aggregated to deliver frequency regulation services for power grids. Although utilizing the idle capacity of battery modules is financially attractive, it remains challenging to consider the heterogeneous module-level characteristics such as dynamic operational efficiencies and battery degradation. In addition, real-time decision making within seconds is required to enable fast frequency response. In order to address these issues, this paper proposes a performance-aware scheduling approach for battery modules to deliver fast frequency response (FFR) support. In particular, the conduction loss and switching loss of battery packs as well as converters are captured within a mix-integer quadratic constrained program (MIQCP). The cycle-based aging model identifies the aging cost of battery modules during frequent cycling by introducing the aging subgradient calculation and linearization. Case studies based on real-world battery data show that the proposed scheduling approach can effectively reduce power loss cost by nearly 28%-57% and battery aging cost by 4%-15% compared to conventional methods, which can also enhance the SoC balance.

eess.SY↗

arXiv2025

Lower Bounds and Accelerated Algorithms in Distributed Stochastic Optimization with Communication Compression

Communication compression is an essential strategy for alleviating communication overhead by reducing the volume of information exchanged between computing nodes in large-scale distributed stochastic optimization. Although numerous algorithms with convergence guarantees have been obtained, the optimal performance limit under communication compression remains unclear. In this paper, we investigate the performance limit of distributed stochastic optimization algorithms employing communication compression. We focus on two main types of compressors, unbiased and contractive, and address the best-possible convergence rates one can obtain with these compressors. We establish the lower bounds for the convergence rates of distributed stochastic optimization in six different settings, combining strongly-convex, generally-convex, or non-convex functions with unbiased or contractive compressor types. To bridge the gap between lower bounds and existing algorithms' rates, we propose NEOLITHIC, a nearly optimal algorithm with compression that achieves the established lower bounds up to logarithmic factors under mild conditions. Extensive experimental results support our theoretical findings. This work provides insights into the theoretical limitations of existing compressors and motivates further research into fundamentally new compressor properties.

cs.LG↗

arXiv2025

CE-LoRA: Computation-Efficient LoRA Fine-Tuning for Language Models

Large Language Models (LLMs) demonstrate exceptional performance across various tasks but demand substantial computational resources even for fine-tuning computation. Although Low-Rank Adaptation (LoRA) significantly alleviates memory consumption during fine-tuning, its impact on computational cost reduction is limited. This paper identifies the computation of activation gradients as the primary bottleneck in LoRA's backward propagation and introduces the Computation-Efficient LoRA (CE-LoRA) algorithm, which enhances computational efficiency while preserving memory efficiency. CE-LoRA leverages two key techniques: Approximated Matrix Multiplication, which replaces dense multiplications of large and complete matrices with sparse multiplications involving only critical rows and columns, and the Double-LoRA technique, which reduces error propagation in activation gradients. Theoretically, CE-LoRA converges at the same rate as LoRA, $ \mathcal{O}(1/\sqrt{T}) $, where $T$ is the number of iteartions. Empirical evaluations confirm that CE-LoRA significantly reduces computational costs compared to LoRA without notable performance degradation.

cs.LG↗

arXiv2025

Detecting Linear Breit-Wheeler Signals with a Laser-Foil Setup

As a fundamental QED process, linear Breit-Wheeler (LBW) pair production predicted 90 years ago has not yet been demonstrated in experiments with real photons. Here, we propose an experimentally advantageous scheme to detect the LBW signal by irradiating a foil target with a single 10 PW-level laser. Our integrated QED particle-in-cell simulations demonstrate that the LBW signal can be explicitly distinguished from the Bethe-Heitler (BH) signal by comparing positron energy spectra behind the target at varying target thicknesses. The LBW positrons are created at the front of the target and subsequently experience both laser vacuum acceleration and sheath field acceleration to gain high energies, while BH positrons, originating within the target bulk, are only subjected to sheath field acceleration. As a result, the invariance of the high-energy tail of positron spectra with respect to the target thickness serves as a distinct signature of the LBW process. Notably, this scheme remains viable even when the BH yield dominates over the LBW yield.

physics.plasm-ph↗

arXiv2024

Ultra-High-Efficiency Dual-Band Thin-Film Lithium Niobate Modulator Incorporating Low-k Underfill with 220 GHz Extrapolated Bandwidth for 390 Gbit/s PAM8 Transmission

High-performance electro-optic modulators play a critical role in modern telecommunication networks and intra-datacenter interconnects. Low driving voltage, large electro-optic bandwidth, compact device size, and multi-band operation ability are essential for various application scenarios, especially energy-efficient high-speed data transmission. However, it is challenging to meet all these requirements simultaneously. Here, we demonstrate a high-performance dual-band thin-film lithium niobate electro-optic modulator with low-k underfill to achieve overall performance improvement. The low-k material helps reduce the RF loss of the modulator and achieve perfect velocity matching with narrow electrode gap to overcome the voltage-bandwidth limitation, extending electro-optic bandwidth and enhancing modulation efficiency simultaneously. The fabricated 7-mm-long modulator exhibits a low half-wave voltage of 1.9 V at C-band and 1.54 V at O-band, featuring a low half-wave voltage-length product of 1.33 V*cm and 1.08 V*cm, respectively. Meanwhile, the novel design yields an ultra-wide extrapolated 3 dB bandwidth of 220 GHz (218 GHz) in the C-band (O-band). High-speed data transmission in both C- and O-bands using the same device has been demonstrated for the first time by PAM8 with data rates up to 390 Gbit/s, corresponding to a record-low energy consumption of 0.69 fJ/bit for next-generation cost-effective ultra-high-speed optical communications.

physics.optics↗

arXiv2024

A Mathematics-Inspired Learning-to-Optimize Framework for Decentralized Optimization

Most decentralized optimization algorithms are handcrafted. While endowed with strong theoretical guarantees, these algorithms generally target a broad class of problems, thereby not being adaptive or customized to specific problem features. This paper studies data-driven decentralized algorithms trained to exploit problem features to boost convergence. Existing learning-to-optimize methods typically suffer from poor generalization or prohibitively vast search spaces. In addition, the vast search space of communicating choices and final goal to reach the global solution via limited neighboring communication cast more challenges in decentralized settings. To resolve these challenges, this paper first derives the necessary conditions that successful decentralized algorithmic rules need to satisfy to achieve both optimality and consensus. Based on these conditions, we propose a novel Mathematics-inspired Learning-to-optimize framework for Decentralized optimization (MiLoDo). Empirical results demonstrate that MiLoDo-trained algorithms outperform handcrafted algorithms and exhibit strong generalizations. Algorithms learned via MiLoDo in 100 iterations perform robustly when running 100,000 iterations during inferences. Moreover, MiLoDo-trained algorithms on synthetic datasets perform well on problems involving real data, higher dimensions, and different loss functions.

math.OC↗

arXiv2024

PianoBART: Symbolic Piano Music Generation and Understanding with Large-Scale Pre-Training

Learning musical structures and composition patterns is necessary for both music generation and understanding, but current methods do not make uniform use of learned features to generate and comprehend music simultaneously. In this paper, we propose PianoBART, a pre-trained model that uses BART for both symbolic piano music generation and understanding. We devise a multi-level object selection strategy for different pre-training tasks of PianoBART, which can prevent information leakage or loss and enhance learning ability. The musical semantics captured in pre-training are fine-tuned for music generation and understanding tasks. Experiments demonstrate that PianoBART efficiently learns musical patterns and achieves outstanding performance in generating high-quality coherent pieces and comprehending music. Our code and supplementary material are available at https://github.com/RS2002/PianoBart.

cs.SD↗

arXiv2024

Distributed Bilevel Optimization with Communication Compression

Stochastic bilevel optimization tackles challenges involving nested optimization structures. Its fast-growing scale nowadays necessitates efficient distributed algorithms. In conventional distributed bilevel methods, each worker must transmit full-dimensional stochastic gradients to the server every iteration, leading to significant communication overhead and thus hindering efficiency and scalability. To resolve this issue, we introduce the first family of distributed bilevel algorithms with communication compression. The primary challenge in algorithmic development is mitigating bias in hypergradient estimation caused by the nested structure. We first propose C-SOBA, a simple yet effective approach with unbiased compression and provable linear speedup convergence. However, it relies on strong assumptions on bounded gradients. To address this limitation, we explore the use of moving average, error feedback, and multi-step compression in bilevel optimization, resulting in a series of advanced algorithms with relaxed assumptions and improved convergence properties. Numerical experiments show that our compressed bilevel algorithms can achieve $10\times$ reduction in communication overhead without severe performance degradation.

math.OC↗

arXiv2024

Inverse Gertsenshtein effect as a probe of high-frequency gravitational waves

We apply the inverse Gertsenshtein effect, i.e., the graviton-photon conversion in the presence of a magnetic field, to constrain high-frequency gravitational waves (HFGWs). Using existing astrophysical measurements, we compute upper limits on the GW energy densities $Ω_{\rm GW}$ at 16 different frequency bands. Given the observed magnetisation of galaxy clusters with field strength $B\simμ{\rm G}$ correlated on $\mathcal{O}(10)\,{\rm kpc}$ scales, we estimate HFGW constraints in the $\mathcal{O}(10^2)\,{\rm GHz}$ regime to be $Ω_{\rm GW}\lesssim10^{16}$ with the temperature measurements of the Atacama Cosmology Telescope (ACT). Similarly, we conservatively obtain $Ω_{\rm GW}\lesssim10^{13} (10^{11})$ in the $\mathcal{O}(10^2)\,{\rm MHz}$ ($\mathcal{O}(10)\,{\rm GHz}$) regime by assuming uniform magnetic field with strength $B\sim0.1\,{\rm nG}$ and saturating the excess signal over the Cosmic Microwave Background (CMB) reported by radio telescopes such as the Experiment to Detect the Global EoR Signature (EDGES), LOw Frequency ARray (LOFAR), and Murchison Widefield Array (MWA), and the balloon-borne second generation Absolute Radiometer for Cosmology, Astrophysics, and Diffuse Emission (ARCADE2) with graviton-induced photons. The upcoming Square Kilometer Array (SKA) can tighten these constraints by roughly 10 orders of magnitude, which will be a step closer to reaching the critical value of $Ω_{\rm GW} = 1$ or the Big Bang Nucleosynthesis (BBN) bound of $Ω_{\rm GW}\simeq1.2\times10^{-6}$. We point to future improvement of the SKA forecast and estimate that proposed CMB measurement at the level of $\mathcal{O}(10^{0-2})\,{\rm nK}$, such as Primordial Inflation Explorer (PIXIE) and Voyage 2050, are needed to viably detect stochastic backgrounds of HFGWs.

astro-ph.CO↗

arXiv2024

Consistency Trajectory Models: Learning Probability Flow ODE Trajectory of Diffusion

Consistency Models (CM) (Song et al., 2023) accelerate score-based diffusion model sampling at the cost of sample quality but lack a natural way to trade-off quality for speed. To address this limitation, we propose Consistency Trajectory Model (CTM), a generalization encompassing CM and score-based models as special cases. CTM trains a single neural network that can -- in a single forward pass -- output scores (i.e., gradients of log-density) and enables unrestricted traversal between any initial and final time along the Probability Flow Ordinary Differential Equation (ODE) in a diffusion process. CTM enables the efficient combination of adversarial training and denoising score matching loss to enhance performance and achieves new state-of-the-art FIDs for single-step diffusion model sampling on CIFAR-10 (FID 1.73) and ImageNet at 64x64 resolution (FID 1.92). CTM also enables a new family of sampling schemes, both deterministic and stochastic, involving long jumps along the ODE solution trajectories. It consistently improves sample quality as computational budgets increase, avoiding the degradation seen in CM. Furthermore, unlike CM, CTM's access to the score function can streamline the adoption of established controllable/conditional generation methods from the diffusion community. This access also enables the computation of likelihood. The code is available at https://github.com/sony/ctm.

cs.LG↗

arXiv2024

Giant enhancement of higher-order harmonics of an optical-tweezer phonon laser

Phonon lasers, as mechanical analogues of optical lasers, are unique tools for not only fundamental studies of phononics but also diverse applications such as acoustic imaging and force sensing. Very recently, by levitating a micro-size sphere in an optical tweezer, higher-order mechanical harmonics were observed in the phonon-lasing regime, as the first step towards nonlinear levitated optomechanics [Nat. Phys. 19, 414 (2023)]. However, both the lasing strengths and the quality factors of the observed harmonics are typically very low, thus severely hindering their applications. Here we show that, by applying a simple but powerful electronic control to such a levitated micro-sphere, three orders of magnitude enhancement are achievable in the brightness of the phonon lasers, including both the fundamental mode and all its higher-order harmonics. Also, giant improvements of their linewidth and frequency stability are realized in such an electro-optomechanical system, together with further improved higher-order phonon coherence. These results, as a significant step forward for enhancing and controlling micro-object phonon lasers, can be readily used for a wide range of applications involving nonlinear phonon lasers, such as acoustic frequency comb, ultra-sound sensing, atmospherical monitoring, and even bio-medical diagnosis of levitated micro-size objects.

physics.optics↗