arXiv subjects

Shi Hu

Publications and source records attributed to Shi Hu.

At least 19 recordsLinked to original sources

arXiv2026

MM-IssueLoc: A Controlled Benchmark for Evaluating Visual Evidence in Multimodal Repository-Level Issue Localization

Real repository issues routinely include visual evidence such as screenshots, error dialogs, rendered UI states, and logs, yet repository-level issue localization is evaluated mostly as a text-only task. Existing multimodal SE benchmarks evaluate end-to-end repair, entangling localization with patch synthesis and obscuring whether visual input helped, hurt, or was ignored. We introduce \textbf{MM-IssueLoc}, a controlled benchmark and evaluation protocol for repository-level localization with visual evidence. MM-IssueLoc contains 652 issue-PR instances across 23 languages, with annotations for 7 image categories and 4 relevance levels. It provides file-level and function-level gold labels, paired text-only and with-image evaluation, and VCE-based diagnostics that convert images into structured textual evidence. We evaluate LLM-based and retrieval-based systems, including MM-IssueLoc-VL-Emb as a controlled multimodal retriever. Results show that existing systems remain far from reliable multimodal repository localization: the strongest agent reaches 38.96 file Acc@5 and 22.45 function Acc@10, while the strongest retriever reaches 33.86 function Acc@10. Cross-benchmark comparisons show that high localization scores on text-dominant SWE benchmarks do not transfer cleanly to multimodal issue localization. MM-IssueLoc turns visual evidence into an explicit evaluation variable, enabling future work to test whether systems improve by using visual evidence for localization, rather than by relying on text-only cues or downstream patch-generation effects.

cs.SE↗

arXiv2026

JoyAI-LLM Flash: Advancing Mid-Scale LLMs with Token Efficiency

We introduce JoyAI-LLM Flash, an efficient Mixture-of-Experts (MoE) language model designed to redefine the trade-off between strong performance and token efficiency in the sub-50B parameter regime. JoyAI-LLM Flash is pretrained on a massive corpus of 20 trillion tokens and further optimized through a rigorous post-training pipeline, including supervised fine-tuning (SFT), Direct Preference Optimization (DPO), and large-scale reinforcement learning (RL) across diverse environments. To improve token efficiency, JoyAI-LLM Flash strategically balances \emph{thinking} and \emph{non-thinking} cognitive modes and introduces FiberPO, a novel RL algorithm inspired by fibration theory that decomposes trust-region maintenance into global and local components, providing unified multi-scale stability control for LLM policy optimization. To enhance architectural sparsity, the model comprises 48B total parameters while activating only 2.7B parameters per forward pass, achieving a substantially higher sparsity ratio than contemporary industry leading models of comparable scale. To further improve inference throughput, we adopt a joint training-inference co-design that incorporates dense Multi-Token Prediction (MTP) and Quantization-Aware Training (QAT). We release the checkpoints for both JoyAI-LLM-48B-A3B Base and its post-trained variants on Hugging Face to support the open-source community.

cs.CL↗

arXiv2026

Falcon-H1R: Pushing the Reasoning Frontiers with a Hybrid Model for Efficient Test-Time Scaling

This work introduces Falcon-H1R, a 7B-parameter reasoning-optimized model that establishes the feasibility of achieving competitive reasoning performance with small language models (SLMs). Falcon-H1R stands out for its parameter efficiency, consistently matching or outperforming SOTA reasoning models that are $2\times$ to $7\times$ larger across a variety of reasoning-intensive benchmarks. These results underscore the importance of careful data curation and targeted training strategies (via both efficient SFT and RL scaling) in delivering significant performance gains without increasing model size. Furthermore, Falcon-H1R advances the 3D limits of reasoning efficiency by combining faster inference (through its hybrid-parallel architecture design), token efficiency, and higher accuracy. This unique blend makes Falcon-H1R-7B a practical backbone for scaling advanced reasoning systems, particularly in scenarios requiring extensive chain-of-thoughts generation and parallel test-time scaling. Leveraging the recently introduced DeepConf approach, Falcon-H1R achieves state-of-the-art test-time scaling efficiency, offering substantial improvements in both accuracy and computational cost. As a result, Falcon-H1R demonstrates that compact models, through targeted model training and architectural choices, can deliver robust and scalable reasoning performance.

cs.AI↗

arXiv2025

Non-Abelian interference of topological edge states

Topological boundary states exhibit distinctive properties, including unidirectional propagation and noise robustness, which hold significant potential for advancing the performance of quantum science and technology. Here, we demonstrate the implementation of non-Abelian quantum interference and entanglement generation, protected by dual symmetries (time-independent inversion and time-dependent interchain), in coupled Su-Schrieffer-Heeger chains. Specifically, in a multichain system, we first achieve tunable topological transfer of a single particle, where the destination chain is selected by the permutation sequence. We then extend this to two particles, observing a non-Abelian Hong-Ou-Mandel interference that generates spatially entangled NOON states whose properties are dictated by the permutation sequence. Our work establishes an alternative pathway for exploring non-Abelian topology applied to quantum science and technology, enabled by the unique protection of time-dependent symmetry.

quant-ph↗

arXiv2025

Falcon-H1: A Family of Hybrid-Head Language Models Redefining Efficiency and Performance

In this report, we introduce Falcon-H1, a new series of large language models (LLMs) featuring hybrid architecture designs optimized for both high performance and efficiency across diverse use cases. Unlike earlier Falcon models built solely on Transformer or Mamba architectures, Falcon-H1 adopts a parallel hybrid approach that combines Transformer-based attention with State Space Models (SSMs), known for superior long-context memory and computational efficiency. We systematically revisited model design, data strategy, and training dynamics, challenging conventional practices in the field. Falcon-H1 is released in multiple configurations, including base and instruction-tuned variants at 0.5B, 1.5B, 1.5B-deep, 3B, 7B, and 34B parameters. Quantized instruction-tuned models are also available, totaling over 30 checkpoints on Hugging Face Hub. Falcon-H1 models demonstrate state-of-the-art performance and exceptional parameter and training efficiency. The flagship Falcon-H1-34B matches or outperforms models up to 70B scale, such as Qwen3-32B, Qwen2.5-72B, and Llama3.3-70B, while using fewer parameters and less data. Smaller models show similar trends: the Falcon-H1-1.5B-Deep rivals current leading 7B-10B models, and Falcon-H1-0.5B performs comparably to typical 7B models from 2024. These models excel across reasoning, mathematics, multilingual tasks, instruction following, and scientific knowledge. With support for up to 256K context tokens and 18 languages, Falcon-H1 is suitable for a wide range of applications. All models are released under a permissive open-source license, underscoring our commitment to accessible and impactful AI research.

cs.CL↗

arXiv2024

Symmetry-protected Landau-Zener-St\"uckelberg-Majorana interference and non-adiabatic topological transport of edge states

We systematically investigate Landau-Zener-St\"uckelberg-Majorana (LZSM) interference under chiral-mirror-like symmetry and propose its application to non-adiabatic topological transport of edge states. Protected by this symmetry, complete destructive interference emerges and can be characterized through occupation probability. This symmetry-protected LZSM interference enables state transitions to be achieved within remarkably short time scales. To demonstrate our mechanism, we provide two distinctive two-level systems as examples and survey them in detail. By tuning evolution speed or increasing holding time, the complete destructive interferences are observed. Furthermore, we make use of this mechanism for topological edge states of Su-Schrieffer-Heeger (SSH) chain by taking them as an isolated two-level system. Through carefully designed time sequences, we construct symmetry-protected LZSM interference of topological edge states, enabling non-adiabatic topological transport. Our work unveils an alternative way to study quantum control, quantum state transfer, and quantum communication via non-adiabatic topological transport.

quant-ph↗

arXiv2024

Hong-Ou-Mandel Interference in a temporal-average-inversion-symmetric chain

We show how to implement tunable beam splitter and Hong-Ou-Mandel interference in the Su-Schrieffer-Heeger chain by manipulating the topological edge states adiabatically. The boson initially injected in the one end of the chain can be transferred to the two-end with a tunable proportion depends on the dynamical phases accumulated during the adiabatic evolution. We also observe Hong-Ou-Mandel interference via the tunable beam splitter ($50:50$) and achieve a spatially entangled two-particle NOON state. We demonstrate the robustness of our proposal under chiral- and time-reversal-symmetry-preserving disorder. However, the chiral symmetry is scarce for realist system. Therefore, we demonstrate Hong-Ou-Mandel interference are robust to inversion symmetric disorder breaking the chiral symmetry, highlighting the protection of inversion symmetry. More importantly, the inversion symmetry violated by static disorder can be restored for more common situations where disorder becomes time dependent, giving rise to the temporal-average-inversion-symmetry protected Hong-Ou-Mandel interference. Our approach opens a new way to study quantum effects in topological matter with potential applications.

quant-ph↗

arXiv2022

Correlated topological pumping of interacting bosons assisted by Bloch oscillations

Thouless pumping, not only achieving quantized transport but also immune to moderate disorder, has attracted growing attention in both experiments and theories. Here, we explore how particle-particle interactions affect topological transport in a periodically-modulated and tilted optical lattice. Not limited to wannier states, our scheme ensures a dispersionless quantized transport even for initial Gaussian-like wave packets of interacting bosons which do not uniformly occupy a given band. This is because the tilting potential leads to Bloch oscillations uniformly sampling the Berry curvatures over the entire Brillouin zone. The interplay among on-site potential difference, tunneling rate and interactions contributes to the topological transport of bound and scattering states and the topologically resonant tunnelings. Our study deepens the understanding of correlation effects on topological states, and provides a feasible way for detecting topological properties in interacting systems.

quant-ph↗

arXiv2022

Adversarial Defense via Image Denoising with Chaotic Encryption

In the literature on adversarial examples, white box and black box attacks have received the most attention. The adversary is assumed to have either full (white) or no (black) access to the defender's model. In this work, we focus on the equally practical gray box setting, assuming an attacker has partial information. We propose a novel defense that assumes everything but a private key will be made available to the attacker. Our framework uses an image denoising procedure coupled with encryption via a discretized Baker map. Extensive testing against adversarial images (e.g. FGSM, PGD) crafted using various gradients shows that our defense achieves significantly better results on CIFAR-10 and CIFAR-100 than the state-of-the-art gray box defenses in both natural and adversarial accuracy.

cs.LG↗

arXiv2021

Uncover band topology via quantized drift in two-dimensional Bloch oscillations

We propose to measure band topology via quantized drift of Bloch oscillations in a two-dimensional Harper-Hofstadter lattice subjected to tilted fields in both directions. When the difference between the two tilted fields is large, Bloch oscillations uniformly sample all momenta, and hence the displacement in each direction tends to be quantized at multiples of the overall period, regardless of any momentum of initial state. The quantized displacement is related to a reduced Chern number defined as a line integral of Berry curvature in each direction, providing an almost perfect measurement of Chern number. Our scheme can apply to detect Chern number and topological phase transitions not only for the energy-separable band, but also for energy-inseparable bands which cannot be achieved by conventional Thouless pumping or integer quantum Hall effect.

cond-mat.quant-gas↗

arXiv2020

Topological pumping assisted by Bloch oscillations

Adiabatic quantum pumping in one-dimensional lattices is extended by adding a tilted potential to probe better topologically nontrivial bands. This extension leads to almost perfectly quantized pumping for an arbitrary initial state selected in a band of interest, including Bloch states. In this approach, the time variable offers not only a synthetic dimension as in the case of the Thouless pumping, but it assists also in the uniform sampling of all momenta due to the Bloch oscillations induced by the tilt. The quantized drift of Bloch oscillations is determined by a one-dimensional time integral of the Berry curvature, being effectively an integer multiple of the topological Chern number in the Thouless pumping. Our study offers a straightforward approach to yield quantized pumping, and it is useful for probing topological phase transitions.

quant-ph↗

arXiv2020

Topological quantum transport and spatial entanglement distribution via disordered bulk channel

We show how to implement quantum transport, generate entangled state and achieve spatial entanglement distribution via topological Thouless pumping in one-dimensional disordered lattices. We introduce the on-site disorders to suppress the high-order resonant tunneling (which cause wavepacket dispersion) and realize dispersionless Thouless pumping. The interplay between the on-site disorders and the band topology enables robust unidirectional transport. We also demonstrate how to prepare spatially entangled two-particle NOON state via Hong-Ou-Mandel interference assisted by the Thouless pumping. The quantum entanglement can be well preserved during the distribution between spatially separated sites. Our system paves a way to wide applications of topology in quantum information process.

quant-ph↗

arXiv2020

Learning to Predict Error for MRI Reconstruction

In healthcare applications, predictive uncertainty has been used to assess predictive accuracy. In this paper, we demonstrate that predictive uncertainty estimated by the current methods does not highly correlate with prediction error by decomposing the latter into random and systematic errors, and showing that the former is equivalent to the variance of the random error. In addition, we observe that current methods unnecessarily compromise performance by modifying the model and training loss to estimate the target and uncertainty jointly. We show that estimating them separately without modifications improves performance. Following this, we propose a novel method that estimates the target labels and magnitude of the prediction error in two steps. We demonstrate this method on a large-scale MRI reconstruction task, and achieve significantly better results than the state-of-the-art uncertainty estimation methods.

cs.LG↗

arXiv2019

Supervised Uncertainty Quantification for Segmentation with Multiple Annotations

The accurate estimation of predictive uncertainty carries importance in medical scenarios such as lung node segmentation. Unfortunately, most existing works on predictive uncertainty do not return calibrated uncertainty estimates, which could be used in practice. In this work we exploit multi-grader annotation variability as a source of 'groundtruth' aleatoric uncertainty, which can be treated as a target in a supervised learning problem. We combine this groundtruth uncertainty with a Probabilistic U-Net and test on the LIDC-IDRI lung nodule CT dataset and MICCAI2012 prostate MRI dataset. We find that we are able to improve predictive uncertainty estimates. We also find that we can improve sample accuracy and sample diversity. In real-world applications, our method could inform doctors about the confidence of the segmentation results.

cs.LG↗

arXiv2019

Dispersion Suppressed Topological Thouless Pumping

In Thouless pumping, although non-flat band has no effects on the quantization of particle transport, it induces wave-packet dispersion which hinders the practical applications of Thouless pumping. Indeed, we find that the dispersion mainly arises from the dynamical phase difference between individual Bloch states. Here we propose two efficient schemes to suppress the dispersion in Thouless pumping: a re-localization echo protocol and a high-order tunneling suppression protocol. In the re-localization echo protocol, we reverse the Hamiltonian in the second pumping cycle to cancel the dynamical phase difference arising from non-flat band, so that the dispersed wave-packet becomes re-localized. In the high-order tunneling suppression protocol, we modulate the nearest-neighbor tunneling to make the Bloch band more flat and suppress the high-order tunneling which causes wave-packet dispersion. Our study paves a way toward the dispersionless Thouless pumping for practical applications in matter transport, state transfer and quantum communication.

quant-ph↗

arXiv2018

Context Encoding Chest X-rays

Chest X-rays are one of the most commonly used technologies for medical diagnosis. Many deep learning models have been proposed to improve and automate the abnormality detection task on this type of data. In this paper, we propose a different approach based on image inpainting under adversarial training first introduced by Goodfellow et al. We configure the context encoder model for this task and train it over 1.1M 128x128 images from healthy X-rays. The goal of our model is to reconstruct the missing central 64x64 patch. Once the model has learned how to inpaint healthy tissue, we test its performance on images with and without abnormalities. We discuss and motivate our results considering PSNR, MSE and SSIM scores as evaluation metrics. In addition, we conduct a 2AFC observer study showing that in half of the times an expert is unable to distinguish real images from the ones reconstructed using our model. By computing and visualizing the pixel-wise difference between the source and the reconstructed images, we can highlight abnormalities to simplify further detection and classification tasks.

cs.CV↗

arXiv2018

Chest X-ray Inpainting with Deep Generative Models

Generative adversarial networks have been successfully applied to inpainting in natural images. However, the current state-of-the-art models have not yet been widely adopted in the medical imaging domain. In this paper, we investigate the performance of three recently published deep learning based inpainting models: context encoders, semantic image inpainting, and the contextual attention model, applied to chest x-rays, as the chest exam is the most commonly performed radiological procedure. We train these generative models on 1.2M 128 $\times$ 128 patches from 60K healthy x-rays, and learn to predict the center 64 $\times$ 64 region in each patch. We test the models on both the healthy and abnormal radiographs. We evaluate the results by visual inspection and comparing the PSNR scores. The outputs of the models are in most cases highly realistic. We show that the methods have potential to enhance and detect abnormalities. In addition, we perform a 2AFC observer study and show that an experienced human observer performs poorly in detecting inpainted regions, particularly those generated by the contextual attention model.

cs.GR↗

arXiv2016

Multi-qubit non-adiabatic holonomic controlled quantum gates in decoherence-free subspaces

Non-adiabatic holonomic quantum gate in decoherence-free subspaces is of greatly practical importance due to its built-in fault tolerance, coherence stabilization virtues, and short run-time. Here we propose some compact schemes to implement two- and three-qubit controlled unitary quantum gates and Fredkin gate. For the controlled unitary quantum gates, the unitary operator acting on the target qubit is an arbitrary single-qubit gate operation. The controlled quantum gates can be directly implemented using non-adiabatic holonomy in decoherence-free subspaces and the required resource for the decoherence-free subspace encoding is minimal by using only two neighboring physical qubits undergoing collective dephasing to encode a logical qubit.

quant-ph↗