Showing 1–50 of 50 results for author: Bai, A

Compiling WebAssembly Concolic Execution with Staging, Continuations, and Snapshots (Extended Version)

Authors: Dinghong Zhong, Alexander Bai, Mikail Khan, Guannan Wei

Abstract: Concolic execution is a variant of symbolic execution that runs a program simultaneously with concrete and symbolic inputs. It records the symbolic constraints encountered along a concrete execution path, then solves those constraints to generate inputs that explore new paths. Existing concolic engines generally follow one of two implementation strategies: Interpreter-based systems are comparative… ▽ More Concolic execution is a variant of symbolic execution that runs a program simultaneously with concrete and symbolic inputs. It records the symbolic constraints encountered along a concrete execution path, then solves those constraints to generate inputs that explore new paths. Existing concolic engines generally follow one of two implementation strategies: Interpreter-based systems are comparatively simple to build but incur substantial interpretation overhead, while instrumentation-based systems avoid this overhead but typically re-execute the program from the beginning for each new input. In this paper, we develop a new approach that achieves the best of both worlds. Starting from the concrete semantics of the target language, we first develop a definitional concolic interpreter and stage it to compile away interpretation overhead while retaining the simplicity of an interpretation-based implementation. By expressing the staged interpreter in continuation-passing style, we can capture execution snapshots at branch points and resume from them when exploring alternative paths, avoiding repeated execution from the program entry. Because snapshot-reuse can itself incur overhead, we further develop a heuristic that favors snapshot-reuse only when it is expected to be beneficial. We instantiate this approach for WebAssembly and implement it in a new concolic-execution compiler GenWasym. Across 184 benchmarks, GenWasym with staging alone achieves a $29.4\times$ average speedup over the interpreter-based WASP; heuristic snapshot-reuse further increases the speedup to $44.9\times$. △ Less

Submitted 20 August, 2026; v1 submitted 18 August, 2026; originally announced August 2026.

Comments: 29 pages; preprint of paper accepted at OOPSLA 2026

Shieldstral

Authors: Antonia Calvi, Avinash Sooriyarachchi, Giada Pistilli, Guillaume Lample, Maarten Buyl, Maximilian Augustin, Maximilian Müller, Pierre Stock, Tom Bewley, Wassim Bouaziz, Yimu Pan, Abdelaziz Bounhar, Abhijeet Somani, Aditi Kabra, Adrian Valente, Adrien Petralia, Adrien Sadé, Alan Jeffares, Albert Jiang, Aleksandr Timashov, Alexandre Cahill, Alexandre Gavaudan, Alexandre Laval, Alexandre Sablayrolles, Amélie Héliou , et al. (251 additional authors not shown)

Abstract: We introduce Shieldstral, a 3B-parameter policy-adaptive multimodal safety classifier that matches or outperforms models nearly 7$\times$ its size on text safety benchmarks and sets a new state of the art on multimodal safety classification. Shieldstral formulates content moderation as a binary question-answering task. This simple formulation unifies diverse moderation tasks into a single yes/no p… ▽ More We introduce Shieldstral, a 3B-parameter policy-adaptive multimodal safety classifier that matches or outperforms models nearly 7$\times$ its size on text safety benchmarks and sets a new state of the art on multimodal safety classification. Shieldstral formulates content moderation as a binary question-answering task. This simple formulation unifies diverse moderation tasks into a single yes/no problem, enabling heterogeneous safety datasets with divergent taxonomies to be consolidated under one training framework. We present the data construction recipe, covering curation and generation of approximately 54.1M samples and a fine-grained evaluation set to evaluate policy adaptability. Together, these enable a small adaptive model to match or outperform much larger models. △ Less

Submitted 4 August, 2026; v1 submitted 28 July, 2026; originally announced July 2026.

Robostral Navigate

Authors: Abdelaziz Bounhar, Abhijeet Somani, Aditi Kabra, Adrian Valente, Adrien Petralia, Adrien Sade, Alan Jeffares, Albert Jiang, Aleksandr Timashov, Alexandre Cahill, Alexandre Gavaudan, Alexandre Laval, Alexandre Sablayrolles, Amelie Heliou, Amos You, Andre Jonasson, Andrew Bai, Andrew Ehrenberg, Andrew Zhao, Angele Lenglemetz, Anmol Agarwal, Antonia Calvi, Arata Suzuki, Arjun Majumdar, Arthur Fournier , et al. (251 additional authors not shown)

Abstract: Deploying navigation systems at scale requires a recipe that minimizes sensor assumptions, generalizes across robot embodiments, and trains efficiently. Yet, today's best systems depend on depth sensors, multi-camera rigs, or pre-built maps, limiting the hardware they support and increasing deployment cost. We introduce Robostral Navigate, an 8B vision-language model built around this scalability… ▽ More Deploying navigation systems at scale requires a recipe that minimizes sensor assumptions, generalizes across robot embodiments, and trains efficiently. Yet, today's best systems depend on depth sensors, multi-camera rigs, or pre-built maps, limiting the hardware they support and increasing deployment cost. We introduce Robostral Navigate, an 8B vision-language model built around this scalability objective. The model consumes only a stream of monocular RGB images - the most ubiquitous sensor across robotic platforms and predicts waypoints by pointing to the next target location in the current camera view. Operating purely in image space, rather than robot-specific coordinates, makes the policy naturally robust to changes in camera intrinsics and scene scale, enabling deployment across wheeled, legged, and aerial robots without recalibration. We generate 2.4 million trajectories across 350k simulated scenes to reduce the reliance on real-world data collection and scale easily. We further introduce a prefix-caching training recipe that packs entire episodes into single training sequences, reducing training tokens by 22x and cutting training time from months to days. A tree-based attention mask prevents conditioning on previous ground-truth actions, encouraging visually grounded action prediction, and reinforcement learning is used to further improve exploration and recovery capabilities. On the Room-to-Room and Room-Across-Room in Continuous Environments (R2R-CE and RxR-CE) benchmarks, Robostral Navigate sets a new state of the art. On R2R-CE, it achieves a 77.4% success rate, surpassing the best monocular method by 10.5 points and the strongest depth- or multi-camera system by 5.3 points despite using only a single RGB camera. On RxR-CE, it reaches 75.1% success rate, outperforming all monocular baselines. △ Less

Submitted 31 July, 2026; v1 submitted 22 July, 2026; originally announced July 2026.

Verifying Probabilistic Programs in Rust

Authors: Alexander Y. Bai, Joseph Tassarotti

Abstract: Recent work has developed many techniques for formally verifying probabilistic programs. However, existing verification frameworks for probabilistic programs are restricted to idealized languages designed for verification. As a result, they cannot be used to verify off-the-shelf probabilistic programs written in standard languages. In contrast, for non-probabilistic programs, a number of verificat… ▽ More Recent work has developed many techniques for formally verifying probabilistic programs. However, existing verification frameworks for probabilistic programs are restricted to idealized languages designed for verification. As a result, they cannot be used to verify off-the-shelf probabilistic programs written in standard languages. In contrast, for non-probabilistic programs, a number of verification tools now support verifying realistic code written in widely used languages such as Go, C, and Rust. To verify probabilistic programs written in these languages, it would be useful to be able to reuse, as much as possible, the extensive development work that has gone into such tools. This paper presents Alerus, a framework for verifying probabilistic Rust programs. Alerus is based on Verus, a verification tool for Rust that supports SMT-based automation and separation-logic-inspired reasoning features. Alerus extends Verus with support for probabilistic reasoning while retaining these expressive features. To do so, Alerus uses a lightweight encoding of probabilistic error credits, a form of ghost state for randomized reasoning introduced in the Eris program logic. By deriving an appropriate specification using error credits, Alerus supports verifying the correctness of randomized sampling algorithms. We use this technique to verify several sampling routines for discrete distributions, including samplers for the discrete Gaussian distributions, the alias method, and the fast loaded dice roller. We establish the soundness of our error credit extension by adapting VerusBelt, a recently developed logical relations model of Verus that encodes its features in terms of the Iris separation logic. To do so, we replace the use of Iris's standard weakest precondition in this model with Eris's probabilistic weakest precondition instead. The resulting soundness proof is fully mechanized in Rocq. △ Less

Submitted 13 July, 2026; originally announced July 2026.

Three-dimensional density and air-rock interface reconstruction with muography: Application to the TianQin tunnel

Authors: Songran Qi, Tao Yu, Shihan Zhao, Yunsong Ning, Aiyu Bai, Yu Chen, Yi Yuan, Mingchen Sun, Zhirui Liu, Liang Xian, Hengye Xu, Hao Jiang, Zhichao Wang, Shuhang Zhang, Su Zhan, Jian Tang

Abstract: Muography is a non-invasive imaging technique that uses cosmic-ray muons, commonly divided into transmission (absorption) and scattering muography. For transmission muography, the inversion algorithm critically determines reconstruction quality. However, widely used schemes may produce smearing artifacts when measurement locations are limited and data are sparse. We develop an optimized Metropolis… ▽ More Muography is a non-invasive imaging technique that uses cosmic-ray muons, commonly divided into transmission (absorption) and scattering muography. For transmission muography, the inversion algorithm critically determines reconstruction quality. However, widely used schemes may produce smearing artifacts when measurement locations are limited and data are sparse. We develop an optimized Metropolis--Hastings (M--H) algorithm that mitigates smearing and retrieves sharper, more accurate density distributions without auxiliary data. Additionally, we implement an inverse distance weighting (IDW) approach to reconstruct the air--rock interface from muon measurements. The optimized M--H algorithm is applied in Monte Carlo simulations and applied to field data from the TianQin Tunnel experiment using the MuGrid-v2 detector. The IDW-reconstructed air--rock interface is validated against Light Detection and Ranging (LiDAR) measurements. In simulations, the optimized M--H algorithm improves high-density anomaly detection precision from $42\%$ to $100\%$ at threshold $5.1\,\mathrm{g/cm^3}$, with gains of $6\%$ to $42\%$ across other threshold and low-density scenarios, together with the TianQin Tunnel reconstructions, these results demonstrate the effectiveness of the proposed approach. △ Less

Submitted 2 June, 2026; originally announced June 2026.

Pro-Tensor Network

Authors: Gen Yue, Ansi Bai, Linqian Wu, Tian Lan

Abstract: We introduce the pro-tensor network, a categorification of the tensor network, as a fully rigorous yet graphically transparent framework for studying the collection of many many-body theories, which we dub many-many-body theory. We provide a comprehensive toolbox for the graphical calculations using pro-tensor networks. As applications, we recover the Levin-Wen model as a "uniform" pro-tensor netw… ▽ More We introduce the pro-tensor network, a categorification of the tensor network, as a fully rigorous yet graphically transparent framework for studying the collection of many many-body theories, which we dub many-many-body theory. We provide a comprehensive toolbox for the graphical calculations using pro-tensor networks. As applications, we recover the Levin-Wen model as a "uniform" pro-tensor network and generalize a result of Kitaev and Kong by characterizing particles as modules over promonads. One can also interpret the string-net pro-tensor network as the space of symmetric tensor networks, thus our framework also applies to the study of generalized symmetry and topological holography. Notably, our generalization dispenses with the assumptions of semisimplicity, finiteness, and rigidity, potentially facilitating the exploration of many-body physics beyond these constraints. △ Less

Submitted 19 May, 2026; v1 submitted 7 May, 2026; originally announced May 2026.

Comments: 96 pages, 21 figures

Voxtral TTS

Authors: Mistral-AI, :, Alexander H. Liu, Alexis Tacnet, Andy Ehrenberg, Andy Lo, Chen-Yo Sun, Guillaume Lample, Henry Lagarde, Jean-Malo Delignon, Jaeyoung Kim, John Harvill, Khyathi Raghavi Chandu, Lorenzo Signoretti, Margaret Jennings, Patrick von Platen, Pavankumar Reddy Muddireddy, Rohin Arora, Sanchit Gandhi, Samuel Humeau, Soham Ghosh, Srijan Mishra, Van Phung, Abdelaziz Bounhar, Abhinav Rastogi , et al. (164 additional authors not shown)

Abstract: We introduce Voxtral TTS, an expressive multilingual text-to-speech model that generates natural speech from as little as 3 seconds of reference audio. Voxtral TTS adopts a hybrid architecture that combines auto-regressive generation of semantic speech tokens with flow-matching for acoustic tokens. These tokens are encoded and decoded with Voxtral Codec, a speech tokenizer trained from scratch wit… ▽ More We introduce Voxtral TTS, an expressive multilingual text-to-speech model that generates natural speech from as little as 3 seconds of reference audio. Voxtral TTS adopts a hybrid architecture that combines auto-regressive generation of semantic speech tokens with flow-matching for acoustic tokens. These tokens are encoded and decoded with Voxtral Codec, a speech tokenizer trained from scratch with a hybrid VQ-FSQ quantization scheme. In human evaluations conducted by native speakers, Voxtral TTS is preferred for multilingual voice cloning due to its naturalness and expressivity, achieving a 68.4\% win rate over ElevenLabs Flash v2.5. We release the model weights under a CC BY-NC license. △ Less

Submitted 6 April, 2026; v1 submitted 26 March, 2026; originally announced March 2026.

Voxtral Realtime

Authors: Mistral-AI, :, Alexander H. Liu, Andy Ehrenberg, Andy Lo, Chen-Yo Sun, Guillaume Lample, Jean-Malo Delignon, Khyathi Raghavi Chandu, Patrick von Platen, Pavankumar Reddy Muddireddy, Rohin Arora, Sanchit Gandhi, Sandeep Subramanian, Soham Ghosh, Srijan Mishra, Abhinav Rastogi, Adrien Sadé, Alan Jeffares, Albert Jiang, Alexandre Cahill, Alexandre Gavaudan, Alexandre Sablayrolles, Amélie Héliou, Amos You , et al. (144 additional authors not shown)

Abstract: We introduce Voxtral Realtime, a natively streaming automatic speech recognition model that matches offline transcription quality at sub-second latency. Unlike approaches that adapt offline models through chunking or sliding windows, Voxtral Realtime is trained end-to-end for streaming, with explicit alignment between audio and text streams. Our architecture builds on the Delayed Streams Modeling… ▽ More We introduce Voxtral Realtime, a natively streaming automatic speech recognition model that matches offline transcription quality at sub-second latency. Unlike approaches that adapt offline models through chunking or sliding windows, Voxtral Realtime is trained end-to-end for streaming, with explicit alignment between audio and text streams. Our architecture builds on the Delayed Streams Modeling framework, introducing a new causal audio encoder and Ada RMS-Norm for improved delay conditioning. We scale pretraining to a large-scale dataset spanning 13 languages. At a delay of 480ms, Voxtral Realtime achieves performance on par with Whisper, the most widely deployed offline transcription system. We release the model weights under the Apache 2.0 license. △ Less

Submitted 6 April, 2026; v1 submitted 11 February, 2026; originally announced February 2026.

Reward-Forcing: Autoregressive Video Generation with Reward Feedback

Authors: Jingran Zhang, Ning Li, Yuanhao Ban, Andrew Bai, Justin Cui

Abstract: While most prior work in video generation relies on bidirectional architectures, recent efforts have sought to adapt these models into autoregressive variants to support near real-time generation. However, such adaptations often depend heavily on teacher models, which can limit performance, particularly in the absence of a strong autoregressive teacher, resulting in output quality that typically l… ▽ More While most prior work in video generation relies on bidirectional architectures, recent efforts have sought to adapt these models into autoregressive variants to support near real-time generation. However, such adaptations often depend heavily on teacher models, which can limit performance, particularly in the absence of a strong autoregressive teacher, resulting in output quality that typically lags behind their bidirectional counterparts. In this paper, we explore an alternative approach that uses reward signals to guide the generation process, enabling more efficient and scalable autoregressive generation. By using reward signals to guide the model, our method simplifies training while preserving high visual fidelity and temporal consistency. Through extensive experiments on standard benchmarks, we find that our approach performs comparably to existing autoregressive models and, in some cases, surpasses similarly sized bidirectional models by avoiding constraints imposed by teacher architectures. For example, on VBench, our method achieves a total score of 84.92, closely matching state-of-the-art autoregressive methods that score 84.31 but require significant heterogeneous distillation. △ Less

Submitted 2 April, 2026; v1 submitted 23 January, 2026; originally announced January 2026.

Comments: https://openreview.net/forum?id=K8Qjsxxl7y&noteId=K8Qjsxxl7y

LoL: Longer than Longer, Scaling Video Generation to Hour

Authors: Justin Cui, Jie Wu, Ming Li, Tao Yang, Xiaojie Li, Rui Wang, Andrew Bai, Yuanhao Ban, Cho-Jui Hsieh

Abstract: Recent research in long-form video generation has shifted from bidirectional to autoregressive models, yet these methods commonly suffer from error accumulation and a loss of long-term coherence. While attention sink frames have been introduced to mitigate this performance decay, they often induce a critical failure mode we term sink-collapse: the generated content repeatedly reverts to the sink f… ▽ More Recent research in long-form video generation has shifted from bidirectional to autoregressive models, yet these methods commonly suffer from error accumulation and a loss of long-term coherence. While attention sink frames have been introduced to mitigate this performance decay, they often induce a critical failure mode we term sink-collapse: the generated content repeatedly reverts to the sink frame, resulting in abrupt scene resets and cyclic motion patterns. Our analysis reveals that sink-collapse originates from an inherent conflict between the periodic structure of Rotary Position Embedding (RoPE) and the multi-head attention mechanisms prevalent in current generative models. To address it, we propose a lightweight, training-free approach that effectively suppresses this behavior by introducing multi-head RoPE jitter that breaks inter-head attention homogenization and mitigates long-horizon collapse. Extensive experiments show that our method successfully alleviates sink-collapse while preserving generation quality. To the best of our knowledge, this work achieves the first demonstration of real-time, streaming, and infinite-length video generation with little quality decay. As an illustration of this robustness, we generate continuous videos up to 12 hours in length, which, to our knowledge, is among the longest publicly demonstrated results in streaming video generation. △ Less

Submitted 23 January, 2026; originally announced January 2026.

Comments: preprint

Dark Dimension Right-handed Neutrinos Confronted with Long-Baseline Oscillation Experiments

Authors: Ai-Yu Bai, Auttakit Chatrabhuti, Yin-Yuan Huang, Hiroshi Isono, Jian Tang

Abstract: Right-handed neutrinos are naturally induced by dark extra dimension models and play an essential role in neutrino oscillations. The model parameters can be examined by the long-baseline neutrino oscillation experiments. In this work, we compute the predicted neutrino oscillation spectra within/without extra dimension models and compare them with the experimental data. We find that the neutrino da… ▽ More Right-handed neutrinos are naturally induced by dark extra dimension models and play an essential role in neutrino oscillations. The model parameters can be examined by the long-baseline neutrino oscillation experiments. In this work, we compute the predicted neutrino oscillation spectra within/without extra dimension models and compare them with the experimental data. We find that the neutrino data in the T2K and NOvA experiments are compatible with the standard neutrino oscillation hypothesis. The results set the stringent exclusion limit on the extra dimension model parameters at a high confidence level. The derived constraints on dark dimension right-handed neutrinos are complementary to those results from the collider experiments and cosmological observations. △ Less

Submitted 8 June, 2026; v1 submitted 2 January, 2026; originally announced January 2026.

Comments: 24 pages, 8 figures. v2: published version

The FACTS Leaderboard: A Comprehensive Benchmark for Large Language Model Factuality

Authors: Aileen Cheng, Alon Jacovi, Amir Globerson, Ben Golan, Charles Kwong, Chris Alberti, Connie Tao, Eyal Ben-David, Gaurav Singh Tomar, Lukas Haas, Yonatan Bitton, Adam Bloniarz, Aijun Bai, Andrew Wang, Anfal Siddiqui, Arturo Bajuelos Castillo, Aviel Atias, Chang Liu, Corey Fry, Daniel Balle, Deepanway Ghosal, Doron Kukliansky, Dror Marcus, Elena Gribovskaya, Eran Ofek , et al. (40 additional authors not shown)

Abstract: We introduce The FACTS Leaderboard, an online leaderboard suite and associated set of benchmarks that comprehensively evaluates the ability of language models to generate factually accurate text across diverse scenarios. The suite provides a holistic measure of factuality by aggregating the performance of models on four distinct sub-leaderboards: (1) FACTS Multimodal, which measures the factuality… ▽ More We introduce The FACTS Leaderboard, an online leaderboard suite and associated set of benchmarks that comprehensively evaluates the ability of language models to generate factually accurate text across diverse scenarios. The suite provides a holistic measure of factuality by aggregating the performance of models on four distinct sub-leaderboards: (1) FACTS Multimodal, which measures the factuality of responses to image-based questions; (2) FACTS Parametric, which assesses models' world knowledge by answering closed-book factoid questions from internal parameters; (3) FACTS Search, which evaluates factuality in information-seeking scenarios, where the model must use a search API; and (4) FACTS Grounding (v2), which evaluates whether long-form responses are grounded in provided documents, featuring significantly improved judge models. Each sub-leaderboard employs automated judge models to score model responses, and the final suite score is an average of the four components, designed to provide a robust and balanced assessment of a model's overall factuality. The FACTS Leaderboard Suite will be actively maintained, containing both public and private splits to allow for external participation while guarding its integrity. It can be found at https://www.kaggle.com/benchmarks/google/facts . △ Less

Submitted 11 December, 2025; originally announced December 2025.

When Distance Distracts: Representation Distance Bias in BT-Loss for Reward Models

Authors: Tong Xie, Andrew Bai, Yuanhao Ban, Yunqi Hong, Haoyu Li, Cho-Jui Hsieh

Abstract: Reward models are central to Large Language Model (LLM) alignment within the framework of RLHF. The standard objective used in reward modeling is the Bradley-Terry (BT) loss, which learns from pairwise data consisting of chosen and rejected responses. In this work, we analyze the per-sample gradient of BT-loss and show spurious learning signals due to representation distance. In particular, BT gra… ▽ More Reward models are central to Large Language Model (LLM) alignment within the framework of RLHF. The standard objective used in reward modeling is the Bradley-Terry (BT) loss, which learns from pairwise data consisting of chosen and rejected responses. In this work, we analyze the per-sample gradient of BT-loss and show spurious learning signals due to representation distance. In particular, BT gradient norm scales with two distinct components: (1) prediction error, reflected by the difference in predicted rewards between chosen and rejected responses, and critically, (2) representation distance between the pair measured in the output space of the final layer. While the first term captures the intended training signal, the second term can significantly impact the update magnitude and misalign learning. Specifically, pairs with small representation distance often receive vanishingly weak updates, even when misranked, while pairs with large distance receive disproportionately strong updates. This leads to gradients from large-distance pairs to overshadow those from small-distance pairs, where fine-grained distinctions are especially important. To overcome this limitation, we propose NormBT, an adaptive pair-wise normalization scheme that rescales updates to balance representation-driven effects and focuses learning signals on prediction error. NormBT is a lightweight, drop-in modification to BT loss with negligible overhead. Across various LLM backbones and datasets, NormBT improves reward model performance consistently, with notable gains of over 5% on the Reasoning category of RewardBench, which contains numerous fine-grained pairs. △ Less

Submitted 8 June, 2026; v1 submitted 6 December, 2025; originally announced December 2025.

Comments: ICML 2026

Tunable Automation in Automated Program Verification

Authors: Alexander Y. Bai, Chris Hawblitzel, Andrea Lattuada

Abstract: Automated verification tools based on SMT solvers have made significant progress in verifying complex software systems. However, these tools face a fundamental tension between automation and performance when dealing with quantifier instantiation -- the primary source of incompleteness and verification slowdown in SMT-based verifiers. Tools choose between aggressive quantifier instantiation that pr… ▽ More Automated verification tools based on SMT solvers have made significant progress in verifying complex software systems. However, these tools face a fundamental tension between automation and performance when dealing with quantifier instantiation -- the primary source of incompleteness and verification slowdown in SMT-based verifiers. Tools choose between aggressive quantifier instantiation that provides more automation but longer verification times, or conservative instantiation that responds quickly but may require more manual proof hints. We present a mechanism that enables fine-grained control over the availability of quantified facts in verification contexts, allowing developers to selectively tune the level of automation. Our approach lets library authors provide different pre-defined automation levels while giving end-users the ability to further customize quantifier availability at the module, function, or proof context level. We implement our techniques in Verus, a Rust-based verification tool, and evaluate them on multiple openly available codebases. Our empirical analysis demonstrates the automation-performance tradeoff and that selective quantifier management enables developers to select the appropriate level of automation in different contexts. △ Less

Submitted 3 December, 2025; originally announced December 2025.

Leveraging Language Models for Interpretable Analysis of Narratives in a Large Corpus

Authors: Eric A. Bai, Minling Zhou, Ricardo Henao, Kyle M. Schwing, Lawrence Carin

Abstract: Narratives drive human behavior and lay at the core of geopolitics, but have eluded quantification that would permit measurement of their overlap and evolution. We present an interpretable model that integrates an established bag-of-words (BoW) topical representation and a novel LLM-based question answering (Q&A) narrative model, which share a latent Reproducing Kernel Hilbert Space representation… ▽ More Narratives drive human behavior and lay at the core of geopolitics, but have eluded quantification that would permit measurement of their overlap and evolution. We present an interpretable model that integrates an established bag-of-words (BoW) topical representation and a novel LLM-based question answering (Q&A) narrative model, which share a latent Reproducing Kernel Hilbert Space representation, to quantify written documents. Our approach mitigates the cost, interpretability, and generalization challenges of using a LLM to analyze large corpora without full inference. We derive efficient functional gradient descent updates that are interpretable and structurally analogous to the self-attention mechanism in Transformers. We further introduce an in-context Q&A extrapolation method inspired by Transformer architectures, enabling accurate prediction of Q&A outcomes for unqueried documents. △ Less

Submitted 23 November, 2025; originally announced November 2025.

Uncertainty-Guided Selective Adaptation Enables Cross-Platform Predictive Fluorescence Microscopy

Authors: Kai-Wen K. Yang, Andrew Bai, Alexandra Bermudez, Yunqi Hong, Zoe Latham, Iris Sloan, Michael Liu, Vishrut Goyal, Cho-Jui Hsieh, Neil Y. C. Lin

Abstract: Deep learning is transforming microscopy, yet models often fail when applied to images from new instruments or acquisition settings. Conventional adversarial domain adaptation (ADDA) retrains entire networks, often disrupting learned semantic representations. Here, we overturn this paradigm by showing that adapting only the earliest convolutional layers, while freezing deeper layers, yields reliab… ▽ More Deep learning is transforming microscopy, yet models often fail when applied to images from new instruments or acquisition settings. Conventional adversarial domain adaptation (ADDA) retrains entire networks, often disrupting learned semantic representations. Here, we overturn this paradigm by showing that adapting only the earliest convolutional layers, while freezing deeper layers, yields reliable transfer. Building on this principle, we introduce Subnetwork Image Translation ADDA with automatic depth selection (SIT-ADDA-Auto), a self-configuring framework that integrates shallow-layer adversarial alignment with predictive uncertainty to automatically select adaptation depth without target labels. We demonstrate robustness via multi-metric evaluation, blinded expert assessment, and uncertainty-depth ablations. Across exposure and illumination shifts, cross-instrument transfer, and multiple stains, SIT-ADDA improves reconstruction and downstream segmentation over full-encoder adaptation and non-adversarial baselines, with reduced drift of semantic features. Our results provide a design rule for label-free adaptation in microscopy and a recipe for field settings; the code is publicly available. △ Less

Submitted 14 November, 2025; originally announced November 2025.

Self-Forcing++: Towards Minute-Scale High-Quality Video Generation

Authors: Justin Cui, Jie Wu, Ming Li, Tao Yang, Xiaojie Li, Rui Wang, Andrew Bai, Yuanhao Ban, Cho-Jui Hsieh

Abstract: Diffusion models have revolutionized image and video generation, achieving unprecedented visual quality. However, their reliance on transformer architectures incurs prohibitively high computational costs, particularly when extending generation to long videos. Recent work has explored autoregressive formulations for long video generation, typically by distilling from short-horizon bidirectional tea… ▽ More Diffusion models have revolutionized image and video generation, achieving unprecedented visual quality. However, their reliance on transformer architectures incurs prohibitively high computational costs, particularly when extending generation to long videos. Recent work has explored autoregressive formulations for long video generation, typically by distilling from short-horizon bidirectional teachers. Nevertheless, given that teacher models cannot synthesize long videos, the extrapolation of student models beyond their training horizon often leads to pronounced quality degradation, arising from the compounding of errors within the continuous latent space. In this paper, we propose a simple yet effective approach to mitigate quality degradation in long-horizon video generation without requiring supervision from long-video teachers or retraining on long video datasets. Our approach centers on exploiting the rich knowledge of teacher models to provide guidance for the student model through sampled segments drawn from self-generated long videos. Our method maintains temporal consistency while scaling video length by up to 20x beyond teacher's capability, avoiding common issues such as over-exposure and error-accumulation without recomputing overlapping frames like previous methods. When scaling up the computation, our method shows the capability of generating videos up to 4 minutes and 15 seconds, equivalent to 99.9% of the maximum span supported by our base model's position embedding and more than 50x longer than that of our baseline model. Experiments on standard benchmarks and our proposed improved benchmark demonstrate that our approach substantially outperforms baseline methods in both fidelity and consistency. Our long-horizon videos demo can be found at https://self-forcing-plus-plus.github.io/ △ Less

Submitted 2 October, 2025; originally announced October 2025.

Comments: preprint

Concepts or Skills? Rethinking Instruction Selection for Multi-modal Models

Authors: Andrew Bai, Justin Cui, Ruochen Wang, Cho-Jui Hsieh

Abstract: Vision-language instruction tuning achieves two main purposes: learning visual concepts and learning visual skills. In this paper, we found that vision-language benchmarks fall into the dichotomy of mainly benefiting from training on instructions with similar skills or visual concepts. Inspired by the discovery, we designed a simple targeted training data selection method to optimize the performan… ▽ More Vision-language instruction tuning achieves two main purposes: learning visual concepts and learning visual skills. In this paper, we found that vision-language benchmarks fall into the dichotomy of mainly benefiting from training on instructions with similar skills or visual concepts. Inspired by the discovery, we designed a simple targeted training data selection method to optimize the performance of a given benchmark. We first extract the concepts/skills from the benchmark, determine whether the benchmark predominantly benefits from similar concepts or skills, and finally select instructions with the most matching concepts/skills. Experiments on 10+ benchmarks validate the effectiveness of our targeted data selection method, showing +0.9\% over the best existing baseline averaged over all benchmarks and +1.5\% on the skill-focused subset. Our findings underscore the importance of recognizing the inherent trade-off within instruction selection, which requires balancing the acquisition of conceptual knowledge against visual skill. △ Less

Submitted 14 August, 2025; originally announced August 2025.

Comments: 11 pages, 1 figure

Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities

Authors: Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, Luke Marris, Sam Petulla, Colin Gaffney, Asaf Aharoni, Nathan Lintz, Tiago Cardal Pais, Henrik Jacobsson, Idan Szpektor, Nan-Jiang Jiang, Krishna Haridasan, Ahmed Omran, Nikunj Saunshi, Dara Bahri, Gaurav Mishra, Eric Chu , et al. (3410 additional authors not shown)

Abstract: In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our most capable model yet, achieving SoTA performance on frontier coding and reasoning benchmarks. In addition to its incredible coding and reasoning skills, Gemini 2.5 Pro is a thinking model that excels at multimodal unde… ▽ More In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our most capable model yet, achieving SoTA performance on frontier coding and reasoning benchmarks. In addition to its incredible coding and reasoning skills, Gemini 2.5 Pro is a thinking model that excels at multimodal understanding and it is now able to process up to 3 hours of video content. Its unique combination of long context, multimodal and reasoning capabilities can be combined to unlock new agentic workflows. Gemini 2.5 Flash provides excellent reasoning abilities at a fraction of the compute and latency requirements and Gemini 2.0 Flash and Flash-Lite provide high performance at low latency and cost. Taken together, the Gemini 2.X model generation spans the full Pareto frontier of model capability vs cost, allowing users to explore the boundaries of what is possible with complex agentic problem solving. △ Less

Submitted 19 December, 2025; v1 submitted 7 July, 2025; originally announced July 2025.

Comments: 72 pages, 17 figures

Unlabeled Data Improves Fine-Grained Image Zero-shot Classification with Multimodal LLMs

Authors: Yunqi Hong, Sohyun An, Andrew Bai, Neil Y. C. Lin, Cho-Jui Hsieh

Abstract: Despite Multimodal Large Language Models (MLLMs) showing promising results on general zero-shot image classification tasks, fine-grained image classification remains challenging. It demands precise attention to subtle visual details to distinguish between visually similar subcategories--details that MLLMs may easily overlook without explicit guidance. To address this, we introduce AutoSEP, an iter… ▽ More Despite Multimodal Large Language Models (MLLMs) showing promising results on general zero-shot image classification tasks, fine-grained image classification remains challenging. It demands precise attention to subtle visual details to distinguish between visually similar subcategories--details that MLLMs may easily overlook without explicit guidance. To address this, we introduce AutoSEP, an iterative self-supervised prompt learning framework designed to enhance MLLM fine-grained classification capabilities in a fully unsupervised manner. Our core idea is to leverage unlabeled data to learn a description prompt that guides MLLMs in identifying crucial discriminative features within an image, and boosts classification accuracy. We developed an automatic self-enhancing prompt learning framework called AutoSEP to iteratively improve the description prompt using unlabeled data, based on instance-level classification scoring function. AutoSEP only requires black-box access to MLLMs, eliminating the need for any training or fine-tuning. We evaluate our approach on multiple fine-grained classification datasets. It consistently outperforms other unsupervised baselines, demonstrating the effectiveness of our self-supervised optimization framework. Notably, AutoSEP on average improves 13 percent over standard zero-shot classification and 5 percent over the best-performing baselines. Code is available at: https://github.com/yq-hong/AutoSEP △ Less

Submitted 26 November, 2025; v1 submitted 1 June, 2025; originally announced June 2025.

MuGrid-v2: A novel scintillator detector for multidisciplinary applications

Authors: Tao Yu, Yunsong Ning, Yi Yuan, Shihan Zhao, Songran Qi, Minchen Sun, Yuye Li, Zhirui Liu, Aiyu Bai, Hesheng Liu, Yibo Lin, Geng Tuo, Ting On Chan, Zhou Zhou, Yu Chen, Yu Chen, Jian Tang

Abstract: Muography, traditionally recognized as a potent instrument for imaging the internal structure of gigantic objects, has initialized various interdisciplinary applications. As the financial and labor costs of muography detector development hinder their massive applications, we develop a novel muon detector called MuGrid by coupling a monolithic plastic scintillator with the light guide array in orde… ▽ More Muography, traditionally recognized as a potent instrument for imaging the internal structure of gigantic objects, has initialized various interdisciplinary applications. As the financial and labor costs of muography detector development hinder their massive applications, we develop a novel muon detector called MuGrid by coupling a monolithic plastic scintillator with the light guide array in order to achieve competitive spatial resolution while substantially reducing production costs. For a prototype detector in 30 cm $\times$ 30 cm, the intrinsic spatial resolution has been optimized toward a millimeter scale. An outdoor field muography experiment was conducted to monitor two buildings for validation purposes. The test successfully resolved the geometric influence of architectural features based on the attenuation of muon flux in a good agreement between experimental results and the simulation prediction. △ Less

Submitted 26 May, 2025; originally announced May 2025.

Journal ref: J. Appl. Phys. 138, 024501 (2025)

Cosmic Ray Muon Polarization to Facilitate Atmospheric Neutrino Physics

Authors: Ming-Chen Sun, Shi-Han Zhao, Rui-Xuan Gao, He-Sheng Liu, Ai-Yu Bai, Jian Tang

Abstract: Atmospheric neutrinos (ATNs) offer a paradigm for understanding neutrino properties, while it is critical to quantify uncertainties in flux modeling. Since ATNs are produced simultaneously with cosmic ray muons, precision measurements of cosmic ray muons, including arrival direction, energy spectra, and spin polarization, will help reduce ATN production uncertainties and facilitate atmospheric neu… ▽ More Atmospheric neutrinos (ATNs) offer a paradigm for understanding neutrino properties, while it is critical to quantify uncertainties in flux modeling. Since ATNs are produced simultaneously with cosmic ray muons, precision measurements of cosmic ray muons, including arrival direction, energy spectra, and spin polarization, will help reduce ATN production uncertainties and facilitate atmospheric neutrino physics. This letter proposes using an array strategy to measure the spin polarization of cosmic ray muons, thereby strengthening the emergent synergies between cosmic ray and atmospheric neutrino physics. Constraints on long-standing atmospheric neutrino flux uncertainties at the percentage level in a few-GeV energy range are achievable within one year using a $O(10)~\text{m}^2$ array of Cosmic-Ray muon Spin polarization detectoRs (CRmuSRs). With the resulting reduction in flux uncertainties, oscillation analysis of atmospheric neutrinos in a liquid scintillator detector with an exposure of 1500 $\text{kt}\cdot\text{yr}$ will break the octant degeneracy and achieve the precision measurement of $θ_{23}$ with the uncertainty smaller than $5$° at 3$σ$ confidence level irrespective of the mass ordering. △ Less

Submitted 14 October, 2025; v1 submitted 19 May, 2025; originally announced May 2025.

Comments: 9 pages; 9 figures

Debugging WebAssembly? Put some Whamm on it!

Authors: Elizabeth Gilbert, Matthew Schneider, Zixi An, Suhas Thalanki, Wavid Bowman, Alexander Bai, Ben L. Titzer, Heather Miller

Abstract: Debugging and monitoring programs are integral to engineering and deploying software. Dynamic analyses monitor applications through source code or IR injection, machine code or bytecode rewriting, and virtual machine or direct hardware support. While these techniques are viable within their respective domains, common tooling across techniques is rare, leading to fragmentation of skills, duplicated… ▽ More Debugging and monitoring programs are integral to engineering and deploying software. Dynamic analyses monitor applications through source code or IR injection, machine code or bytecode rewriting, and virtual machine or direct hardware support. While these techniques are viable within their respective domains, common tooling across techniques is rare, leading to fragmentation of skills, duplicated efforts, and inconsistent feature support. We address this problem in the WebAssembly ecosystem with Whamm, a declarative instrumentation DSL for WebAssembly that abstracts above the instrumentation strategy, leveraging bytecode rewriting and engine support as available. Whamm solves three problems: 1) tooling fragmentation, 2) prohibitive instrumentation overhead of general-purpose frameworks, and 3) tedium of tailoring low-level high-performance mechanisms. Whamm provides fully-programmable instrumentation with declarative match rules, static and dynamic predication, automatic state reporting, and user library support, while achieving high performance through compiler and engine optimizations. At the back end, Whamm provides instrumentation to a Wasm engine as Wasm code, reusing existing engine optimizations and unlocking new ones, most notably intrinsification, to minimize overhead. In particular, explicitly requesting program state in match rules, rather than reflection, enables the engine to efficiently bundle arguments and even inline compiled probe logic. Whamm streamlines the tooling effort, as its bytecode-rewriting target can run instrumented programs everywhere, lowering fragmentation and advancing the state of the art for engine support. We evaluate Whamm with case studies of non-trivial monitors and show it is expressive, powerful, and efficient. △ Less

Submitted 28 April, 2025; originally announced April 2025.

Exploring Expert Failures Improves LLM Agent Tuning

Authors: Li-Cheng Lan, Andrew Bai, Minhao Cheng, Cho-Jui Hsieh, Tianyi Zhou

Abstract: Large Language Models (LLMs) have shown tremendous potential as agents, excelling at tasks that require multiple rounds of reasoning and interactions. Rejection Sampling Fine-Tuning (RFT) has emerged as an effective method for finetuning LLMs as agents: it first imitates expert-generated successful trajectories and further improves agentic skills through iterative fine-tuning on successful, self-g… ▽ More Large Language Models (LLMs) have shown tremendous potential as agents, excelling at tasks that require multiple rounds of reasoning and interactions. Rejection Sampling Fine-Tuning (RFT) has emerged as an effective method for finetuning LLMs as agents: it first imitates expert-generated successful trajectories and further improves agentic skills through iterative fine-tuning on successful, self-generated trajectories. However, since the expert (e.g., GPT-4) succeeds primarily on simpler subtasks and RFT inherently favors simpler scenarios, many complex subtasks remain unsolved and persistently out-of-distribution (OOD). Upon investigating these challenging subtasks, we discovered that previously failed expert trajectories can often provide valuable guidance, e.g., plans and key actions, that can significantly improve agent exploration efficiency and acquisition of critical skills. Motivated by these observations, we propose Exploring Expert Failures (EEF), which identifies beneficial actions from failed expert trajectories and integrates them into the training dataset. Potentially harmful actions are meticulously excluded to prevent contamination of the model learning process. By leveraging the beneficial actions in expert failures, EEF successfully solves some previously unsolvable subtasks and improves agent tuning performance. Remarkably, our approach achieved a 62\% win rate in WebShop, outperforming RFT (53. 6\%) and GPT-4 (35. 6\%), and to the best of our knowledge, setting a new state-of-the-art as the first method to surpass a score of 0.81 in WebShop and exceed 81 in SciWorld. △ Less

Submitted 18 April, 2025; v1 submitted 17 April, 2025; originally announced April 2025.

On the Representation Categories of Weak Hopf Algebras Arising from Levin-Wen Models

Authors: Ansi Bai, Zhi-Hao Zhang

Abstract: In their study of Levin-Wen models [Commun. Math. Phys. 313 (2012) 351-373], Kitaev and Kong proposed a weak Hopf algebra associated with a unitary fusion category $\mathcal{C}$ and a unitary left $\mathcal{C}$-module $\mathcal{M}$, and sketched a proof that its representation category is monoidally equivalent to the unitary $\mathcal{C}$-module functor category… ▽ More In their study of Levin-Wen models [Commun. Math. Phys. 313 (2012) 351-373], Kitaev and Kong proposed a weak Hopf algebra associated with a unitary fusion category $\mathcal{C}$ and a unitary left $\mathcal{C}$-module $\mathcal{M}$, and sketched a proof that its representation category is monoidally equivalent to the unitary $\mathcal{C}$-module functor category $\mathrm{Fun}^{\mathrm{u}}_{\mathcal{C}}(\mathcal{M},\mathcal{M})^\mathrm{rev}$. We give an independent proof of this result without the unitarity conditions. In particular, viewing $\mathcal{C}$ as a left $\mathcal{C} \boxtimes \mathcal{C}^{\mathrm{rev}}$-module, we obtain a quasi-triangular weak Hopf algebra whose representation category is braided equivalent to the Drinfeld center $\mathcal{Z}(\mathcal{C})$. In the appendix, we also compare this quasi-triangular weak Hopf algebra with the tube algebra $\mathrm{Tube}_{\mathcal{C}}$ of $\mathcal{C}$ when $\mathcal{C}$ is pivotal. These two algebras are Morita equivalent by the well-known equivalence $\mathrm{Rep}(\mathrm{Tube}_{\mathcal{C}})\cong\mathcal{Z}(\mathcal{C})$. However, we show that in general there is no weak Hopf algebra structure on $\mathrm{Tube}_{\mathcal{C}}$ such that the above equivalence is monoidal. △ Less

Submitted 9 March, 2025; originally announced March 2025.

Comments: 58 pages, 0 figures, with an appendix on Ocneanu's tube algebras

MSC Class: 16T05 (Primary) 18M20; 81R50 (Secondary)

On the Loss of Context-awareness in General Instruction Fine-tuning

Authors: Yihan Wang, Andrew Bai, Nanyun Peng, Cho-Jui Hsieh

Abstract: Pre-trained Large Language Models (LLMs) require post-training methods such as supervised fine-tuning (SFT) on instruction-response pairs to enable instruction following. However, this process can potentially harm existing capabilities learned during pre-training. In this paper, we investigate the loss of context awareness after SFT, where context awareness is defined as the ability to extract and… ▽ More Pre-trained Large Language Models (LLMs) require post-training methods such as supervised fine-tuning (SFT) on instruction-response pairs to enable instruction following. However, this process can potentially harm existing capabilities learned during pre-training. In this paper, we investigate the loss of context awareness after SFT, where context awareness is defined as the ability to extract and understand information from user-provided context and respond accordingly. We identify and demonstrate that the loss of context awareness, particularly in open-source models, occurs in instruction fine-tuned LLMs when the chat template is applied to input prompts. We identify that the performance decline is associated with a bias toward different roles learned during conversational instruction fine-tuning. We demonstrate this correlation by visualizing changes in attention allocation after the chat template is applied and manually steering the attention heads. The bias can be learned from training examples that align with the model's internal knowledge and rely less on the user-provided context to generate correct responses. Based on these observations, we propose a metric to identify context-dependent examples from general instruction fine-tuning datasets. We then apply conditional instruction fine-tuning with a context-dependency indicator, enabling the model to preserve context awareness after SFT. Empirical experiments on four context-dependent downstream tasks and three pre-trained LLMs of different sizes show that our method effectively mitigates the loss of context awareness without compromising general instruction-following capabilities. △ Less

Submitted 2 February, 2025; v1 submitted 4 November, 2024; originally announced November 2024.

Conceptual Design of the Muonium-to-Antimuonium Conversion Experiment (MACE)

Authors: Ai-Yu Bai, Hanjie Cai, Chang-Lin Chen, Siyuan Chen, Xurong Chen, Yu Chen, Weibin Cheng, Ling-Yun Dai, Rui-Rui Fan, Li Gong, Zihao Guo, Yuan He, Zhilong Hou, Yinyuan Huang, Huan Jia, Hao Jiang, Han-Tao Jing, Xiaoshen Kang, Hai-Bo Li, Jincheng Li, Yang Li, Daming Liu, Shulin Liu, Guihao Lu, Han Miao , et al. (27 additional authors not shown)

Abstract: The spontaneous conversion of muonium to antimuonium is one of the interesting charged lepton flavor violation phenomena offering a sensitive probe of potential new physics and serving as a tool to constrain the parameter space beyond the Standard Model. The Muonium-to-Antimuonium Conversion Experiment (MACE) is designed to utilize a high-intensity muon beam, a Michel electron magnetic spectromete… ▽ More The spontaneous conversion of muonium to antimuonium is one of the interesting charged lepton flavor violation phenomena offering a sensitive probe of potential new physics and serving as a tool to constrain the parameter space beyond the Standard Model. The Muonium-to-Antimuonium Conversion Experiment (MACE) is designed to utilize a high-intensity muon beam, a Michel electron magnetic spectrometer, a positron transport system, and a positron detection system, to either discover or constrain this rare process with a conversion probability of $\mathcal{O}(10^{-13})$. This article presents an overview of the theoretical framework as well as a detailed description of the experimental design for the search for muonium-to-antimuonium conversion. △ Less

Submitted 21 November, 2025; v1 submitted 24 October, 2024; originally announced October 2024.

Comments: 45 pages, 51 figures, 14 tables. Accepted by Nuclear Science and Techniques

Journal ref: Nuclear Science and Techniques, volume 37, article number 57, (2026)

FiTv2: Scalable and Improved Flexible Vision Transformer for Diffusion Model

Authors: ZiDong Wang, Zeyu Lu, Di Huang, Cai Zhou, Wanli Ouyang, and Lei Bai

Abstract: \textit{Nature is infinitely resolution-free}. In the context of this reality, existing diffusion models, such as Diffusion Transformers, often face challenges when processing image resolutions outside of their trained domain. To address this limitation, we conceptualize images as sequences of tokens with dynamic sizes, rather than traditional methods that perceive images as fixed-resolution grids… ▽ More \textit{Nature is infinitely resolution-free}. In the context of this reality, existing diffusion models, such as Diffusion Transformers, often face challenges when processing image resolutions outside of their trained domain. To address this limitation, we conceptualize images as sequences of tokens with dynamic sizes, rather than traditional methods that perceive images as fixed-resolution grids. This perspective enables a flexible training strategy that seamlessly accommodates various aspect ratios during both training and inference, thus promoting resolution generalization and eliminating biases introduced by image cropping. On this basis, we present the \textbf{Flexible Vision Transformer} (FiT), a transformer architecture specifically designed for generating images with \textit{unrestricted resolutions and aspect ratios}. We further upgrade the FiT to FiTv2 with several innovative designs, includingthe Query-Key vector normalization, the AdaLN-LoRA module, a rectified flow scheduler, and a Logit-Normal sampler. Enhanced by a meticulously adjusted network structure, FiTv2 exhibits $2\times$ convergence speed of FiT. When incorporating advanced training-free extrapolation techniques, FiTv2 demonstrates remarkable adaptability in both resolution extrapolation and diverse resolution generation. Additionally, our exploration of the scalability of the FiTv2 model reveals that larger models exhibit better computational efficiency. Furthermore, we introduce an efficient post-training strategy to adapt a pre-trained model for the high-resolution generation. Comprehensive experiments demonstrate the exceptional performance of FiTv2 across a broad range of resolutions. We have released all the codes and models at \url{https://github.com/whlzy/FiT} to promote the exploration of diffusion transformer models for arbitrary-resolution image generation. △ Less

Submitted 17 October, 2024; originally announced October 2024.

Comments: arXiv admin note: text overlap with arXiv:2402.12376

Inference Scaling for Long-Context Retrieval Augmented Generation

Authors: Zhenrui Yue, Honglei Zhuang, Aijun Bai, Kai Hui, Rolf Jagerman, Hansi Zeng, Zhen Qin, Dong Wang, Xuanhui Wang, Michael Bendersky

Abstract: The scaling of inference computation has unlocked the potential of long-context large language models (LLMs) across diverse settings. For knowledge-intensive tasks, the increased compute is often allocated to incorporate more external knowledge. However, without effectively utilizing such knowledge, solely expanding context does not always enhance performance. In this work, we investigate inferenc… ▽ More The scaling of inference computation has unlocked the potential of long-context large language models (LLMs) across diverse settings. For knowledge-intensive tasks, the increased compute is often allocated to incorporate more external knowledge. However, without effectively utilizing such knowledge, solely expanding context does not always enhance performance. In this work, we investigate inference scaling for retrieval augmented generation (RAG), exploring the combination of multiple strategies beyond simply increasing the quantity of knowledge, including in-context learning and iterative prompting. These strategies provide additional flexibility to scale test-time computation (e.g., by increasing retrieved documents or generation steps), thereby enhancing LLMs' ability to effectively acquire and utilize contextual information. We address two key questions: (1) How does RAG performance benefit from the scaling of inference computation when optimally configured? (2) Can we predict the optimal test-time compute allocation for a given budget by modeling the relationship between RAG performance and inference parameters? Our observations reveal that increasing inference computation leads to nearly linear gains in RAG performance when optimally allocated, a relationship we describe as the inference scaling laws for RAG. Building on this, we further develop the computation allocation model to estimate RAG performance across different inference configurations. The model predicts optimal inference parameters under various computation constraints, which align closely with the experimental results. By applying these optimal configurations, we demonstrate that scaling inference compute on long-context LLMs achieves up to 58.9% gains on benchmark datasets compared to standard RAG. △ Less

Submitted 2 March, 2025; v1 submitted 5 October, 2024; originally announced October 2024.

Comments: ICLR 2025

CLUE: Concept-Level Uncertainty Estimation for Large Language Models

Authors: Yu-Hsiang Wang, Andrew Bai, Che-Ping Tsai, Cho-Jui Hsieh

Abstract: Large Language Models (LLMs) have demonstrated remarkable proficiency in various natural language generation (NLG) tasks. Previous studies suggest that LLMs' generation process involves uncertainty. However, existing approaches to uncertainty estimation mainly focus on sequence-level uncertainty, overlooking individual pieces of information within sequences. These methods fall short in separately… ▽ More Large Language Models (LLMs) have demonstrated remarkable proficiency in various natural language generation (NLG) tasks. Previous studies suggest that LLMs' generation process involves uncertainty. However, existing approaches to uncertainty estimation mainly focus on sequence-level uncertainty, overlooking individual pieces of information within sequences. These methods fall short in separately assessing the uncertainty of each component in a sequence. In response, we propose a novel framework for Concept-Level Uncertainty Estimation (CLUE) for LLMs. We leverage LLMs to convert output sequences into concept-level representations, breaking down sequences into individual concepts and measuring the uncertainty of each concept separately. We conduct experiments to demonstrate that CLUE can provide more interpretable uncertainty estimation results compared with sentence-level uncertainty, and could be a useful tool for various tasks such as hallucination detection and story generation. △ Less

Submitted 4 September, 2024; originally announced September 2024.

Embedding Space Selection for Detecting Memorization and Fingerprinting in Generative Models

Authors: Jack He, Jianxing Zhao, Andrew Bai, Cho-Jui Hsieh

Abstract: In the rapidly evolving landscape of artificial intelligence, generative models such as Generative Adversarial Networks (GANs) and Diffusion Models have become cornerstone technologies, driving innovation in diverse fields from art creation to healthcare. Despite their potential, these models face the significant challenge of data memorization, which poses risks to privacy and the integrity of gen… ▽ More In the rapidly evolving landscape of artificial intelligence, generative models such as Generative Adversarial Networks (GANs) and Diffusion Models have become cornerstone technologies, driving innovation in diverse fields from art creation to healthcare. Despite their potential, these models face the significant challenge of data memorization, which poses risks to privacy and the integrity of generated content. Among various metrics of memorization detection, our study delves into the memorization scores calculated from encoder layer embeddings, which involves measuring distances between samples in the embedding spaces. Particularly, we find that the memorization scores calculated from layer embeddings of Vision Transformers (ViTs) show an notable trend - the latter (deeper) the layer, the less the memorization measured. It has been found that the memorization scores from the early layers' embeddings are more sensitive to low-level memorization (e.g. colors and simple patterns for an image), while those from the latter layers are more sensitive to high-level memorization (e.g. semantic meaning of an image). We also observe that, for a specific model architecture, its degree of memorization on different levels of information is unique. It can be viewed as an inherent property of the architecture. Building upon this insight, we introduce a unique fingerprinting methodology. This method capitalizes on the unique distributions of the memorization score across different layers of ViTs, providing a novel approach to identifying models involved in generating deepfakes and malicious content. Our approach demonstrates a marked 30% enhancement in identification accuracy over existing baseline methods, offering a more effective tool for combating digital misinformation. △ Less

Submitted 30 July, 2024; originally announced July 2024.

Defending LLMs against Jailbreaking Attacks via Backtranslation

Authors: Yihan Wang, Zhouxing Shi, Andrew Bai, Cho-Jui Hsieh

Abstract: Although many large language models (LLMs) have been trained to refuse harmful requests, they are still vulnerable to jailbreaking attacks which rewrite the original prompt to conceal its harmful intent. In this paper, we propose a new method for defending LLMs against jailbreaking attacks by ``backtranslation''. Specifically, given an initial response generated by the target LLM from an input pro… ▽ More Although many large language models (LLMs) have been trained to refuse harmful requests, they are still vulnerable to jailbreaking attacks which rewrite the original prompt to conceal its harmful intent. In this paper, we propose a new method for defending LLMs against jailbreaking attacks by ``backtranslation''. Specifically, given an initial response generated by the target LLM from an input prompt, our backtranslation prompts a language model to infer an input prompt that can lead to the response. The inferred prompt is called the backtranslated prompt which tends to reveal the actual intent of the original prompt, since it is generated based on the LLM's response and not directly manipulated by the attacker. We then run the target LLM again on the backtranslated prompt, and we refuse the original prompt if the model refuses the backtranslated prompt. We explain that the proposed defense provides several benefits on its effectiveness and efficiency. We empirically demonstrate that our defense significantly outperforms the baselines, in the cases that are hard for the baselines, and our defense also has little impact on the generation quality for benign input prompts. Our implementation is based on our library for LLM jailbreaking defense algorithms at \url{https://github.com/YihanWang617/llm-jailbreaking-defense}, and the code for reproducing our experiments is available at \url{https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation}. △ Less

Submitted 6 June, 2024; v1 submitted 26 February, 2024; originally announced February 2024.

An Efficient Rehearsal Scheme for Catastrophic Forgetting Mitigation during Multi-stage Fine-tuning

Authors: Andrew Bai, Chih-Kuan Yeh, Cho-Jui Hsieh, Ankur Taly

Abstract: Incrementally fine-tuning foundational models on new tasks or domains is now the de facto approach in NLP. A known pitfall of this approach is the \emph{catastrophic forgetting} of prior knowledge that happens during fine-tuning. A common approach to alleviate such forgetting is to rehearse samples from prior tasks during fine-tuning. Several existing works assume a fixed memory buffer to store pr… ▽ More Incrementally fine-tuning foundational models on new tasks or domains is now the de facto approach in NLP. A known pitfall of this approach is the \emph{catastrophic forgetting} of prior knowledge that happens during fine-tuning. A common approach to alleviate such forgetting is to rehearse samples from prior tasks during fine-tuning. Several existing works assume a fixed memory buffer to store prior task examples, while relying on inferences (forward passes) with the model at hand for choosing examples for rehearsal from the buffer. However, given the increasing computational cost of model inference, and decreasing cost of data storage, we focus on the setting to rehearse samples with a fixed computational budget instead of a fixed memory budget. We propose a sampling scheme, \texttt{\bf mix-cd}, that prioritizes rehearsal of ``collateral damage'' samples, which are samples predicted correctly by the prior model but forgotten by the incrementally tuned one. The crux of our scheme is a procedure to efficiently estimate the density of collateral damage samples without incurring additional model inferences. Our approach is computationally efficient, easy to implement, and outperforms several leading continual learning methods in compute-constrained settings. All the code will be publicly available at https://github.com/jybai/mix-cd-rehearsal. △ Less

Submitted 11 February, 2025; v1 submitted 12 February, 2024; originally announced February 2024.

Comments: 13 pages, 9 figures. Published in NAACL 2025 Findings

Data Attribution for Diffusion Models: Timestep-induced Bias in Influence Estimation

Authors: Tong Xie, Haoyu Li, Andrew Bai, Cho-Jui Hsieh

Abstract: Data attribution methods trace model behavior back to its training dataset, offering an effective approach to better understand ''black-box'' neural networks. While prior research has established quantifiable links between model output and training data in diverse settings, interpreting diffusion model outputs in relation to training samples remains underexplored. In particular, diffusion models o… ▽ More Data attribution methods trace model behavior back to its training dataset, offering an effective approach to better understand ''black-box'' neural networks. While prior research has established quantifiable links between model output and training data in diverse settings, interpreting diffusion model outputs in relation to training samples remains underexplored. In particular, diffusion models operate over a sequence of timesteps instead of instantaneous input-output relationships in previous contexts, posing a significant challenge to extend existing frameworks to diffusion models directly. Notably, we present Diffusion-TracIn that incorporates this temporal dynamics and observe that samples' loss gradient norms are highly dependent on timestep. This trend leads to a prominent bias in influence estimation, and is particularly noticeable for samples trained on large-norm-inducing timesteps, causing them to be generally influential. To mitigate this effect, we introduce Diffusion-ReTrac as a re-normalized adaptation that enables the retrieval of training samples more targeted to the test sample of interest, facilitating a localized measurement of influence and considerably more intuitive visualization. We demonstrate the efficacy of our approach through various evaluation metrics and auxiliary tasks, reducing the amount of generally influential samples to $\frac{1}{3}$ of its original quantity. △ Less

Submitted 28 July, 2024; v1 submitted 17 January, 2024; originally announced January 2024.

Comments: Accepted at TMLR. Code available at https://github.com/txie1/diffusion-ReTrac

Optimal $F$-score Clustering for Bipartite Record Linkage

Authors: Eric A. Bai, Olivier Binette, Jerome P. Reiter

Abstract: Probabilistic record linkage is often used to match records from two files, in particular when the variables common to both files comprise imperfectly measured identifiers like names and demographic variables. We consider bipartite record linkage settings in which each entity appears at most once within a file, i.e., there are no duplicates within the files, but some entities appear in both files.… ▽ More Probabilistic record linkage is often used to match records from two files, in particular when the variables common to both files comprise imperfectly measured identifiers like names and demographic variables. We consider bipartite record linkage settings in which each entity appears at most once within a file, i.e., there are no duplicates within the files, but some entities appear in both files. In this setting, the analyst desires a point estimate of the linkage structure that matches each record to at most one record from the other file. We propose an approach for obtaining this point estimate by maximizing the expected $F$-score for the linkage structure. We target the approach for record linkage methods that produce either (an approximate) posterior distribution of the unknown linkage structure or probabilities of matches for record pairs. Using simulations and applications with genuine data, we illustrate that the $F$-score estimators can lead to sensible estimates of the linkage structure. △ Less

Submitted 4 December, 2023; v1 submitted 23 November, 2023; originally announced November 2023.

Beyond the Screen: Safeguarding Mental Health in the Digital Workplace Through Organizational Commitment and Ethical Environment

Authors: Ali Bai, Morteza Vahedian

Abstract: This research explores the intricate relationship between organizational commitment and nomophobia, illuminating the mediating influence of the ethical environment. Utilizing Meyer and Allen's three-component model, the study finds a significant inverse correlation between organizational commitment and nomophobia, highlighting how strong organizational ties can alleviate the anxiety of digital dis… ▽ More This research explores the intricate relationship between organizational commitment and nomophobia, illuminating the mediating influence of the ethical environment. Utilizing Meyer and Allen's three-component model, the study finds a significant inverse correlation between organizational commitment and nomophobia, highlighting how strong organizational ties can alleviate the anxiety of digital disconnection. The ethical environment further emerges as a significant mediator, indicating its dual role in promoting ethical behavior and mitigating nomophobia's psychological effects. The study's theoretical advancement lies in its empirical evidence on the seldom-explored nexus between organizational commitment and technology-induced stress. By integrating organizational ethics and technological impact, the research offers a novel perspective on managing digital dependence in the workplace. From a practical standpoint, this study serves as a catalyst for organizational leaders to reinforce affective and normative commitment, thereby reducing nomophobia. The findings underscore the necessity of ethical leadership and comprehensive ethical policies as foundations for employee well-being in the digital age. Conclusively, this study delineates the protective role of organizational commitment and the significance of ethical environments, guiding organizations to foster cultures that balance technological efficiency with employee welfare. As a contribution to both academic discourse and practical application, it emphasizes the importance of nurturing a supportive and ethically sound workplace in an era of pervasive digital integration. △ Less

Submitted 4 November, 2023; originally announced November 2023.

Elevating Women in the Workplace: The Dual Influence of Spiritual Intelligence and Ethical Environments on Job Satisfaction

Authors: Ali Bai, Morteza Vahedian, Rashin Ghahreman, Hasan Piri

Abstract: In today's rapidly evolving workplace, the dynamics of job satisfaction and its determinants have become a focal point of organizational studies. This research offers a comprehensive examination of the nexus between spiritual intelligence and job satisfaction among female employees, with particular emphasis on the moderating role of ethical work environments. Beginning with an exploration of the m… ▽ More In today's rapidly evolving workplace, the dynamics of job satisfaction and its determinants have become a focal point of organizational studies. This research offers a comprehensive examination of the nexus between spiritual intelligence and job satisfaction among female employees, with particular emphasis on the moderating role of ethical work environments. Beginning with an exploration of the multifaceted nature of human needs, the study delves deep into the psychological underpinnings that drive job satisfaction. It elucidates how various tangible and intangible motivators, such as salary benefits and recognition, play pivotal roles in shaping employee attitudes and behaviors. Moreover, the research spotlights the unique challenges and experiences of female employees, advocating for a more inclusive understanding of their needs. An extensive review of the literature and empirical analysis culminates in the pivotal finding that integrating spiritual intelligence and ethical considerations within organizational practices can significantly enhance job satisfaction. Such a holistic approach, the paper posits, not only bolsters the well-being and contentment of female employees but also augments overall organizational productivity, retention rates, and morale. △ Less

Submitted 24 October, 2023; originally announced October 2023.

Research Note: Bayesian Record Linkage with Application to Chinese Immigrants in Raleigh-Durham (ChIRDU) Study

Authors: Eric A. Bai, Madeleine Beckner, Botao Ju, Jerome P. Reiter, Ted Mouw, M. Giovanna Merli

Abstract: Many population surveys do not provide information on respondents' residential addresses, instead offering coarse geographies like zip code or higher aggregations. However, fine resolution geography can be beneficial for characterizing neighborhoods, especially for relatively rare populations such as immigrants. One way to obtain such information is to link survey records to records in auxiliary d… ▽ More Many population surveys do not provide information on respondents' residential addresses, instead offering coarse geographies like zip code or higher aggregations. However, fine resolution geography can be beneficial for characterizing neighborhoods, especially for relatively rare populations such as immigrants. One way to obtain such information is to link survey records to records in auxiliary databases that include residential addresses by matching on variables common to both files. In this research note, we present an approach based on probabilistic record linkage that enables matching survey participants in the Chinese Immigrants in Raleigh-Durham (ChIRDU) Study to records from InfoUSA, an information provider of residential records. The two files use different Chinese name romanization practices, which we address through a novel and generalizable strategy for constructing records' pairwise comparison vectors for romanized names. Using a fully Bayesian record linkage model, we characterize the geospatial distribution of Chinese immigrants in the Raleigh-Durham area. △ Less

Submitted 21 October, 2023; originally announced October 2023.

Examining the Influence of Job Satisfaction on Individual Innovation and Its Components: Considering the Moderating Role of Technostress

Authors: Fatemeh Daneshmandi, Hassan Hessari, Tahmineh Nategh, Ali Bai

Abstract: Background: Employee innovation is a crucial aspect of organizations in the current era. Therefore, studying the factors influencing individual innovation is vital and unavoidable. Undoubtedly, job satisfaction is a significant variable in management sciences. Nowadays, all organizations are interconnected with technology. Objective: This research explores the relationship between job satisfaction… ▽ More Background: Employee innovation is a crucial aspect of organizations in the current era. Therefore, studying the factors influencing individual innovation is vital and unavoidable. Undoubtedly, job satisfaction is a significant variable in management sciences. Nowadays, all organizations are interconnected with technology. Objective: This research explores the relationship between job satisfaction and individual innovation, including its components, and the moderating role of technostress. Research Method: This study, in terms of purpose, is applied, and in terms of data collection method, it is a descriptive survey. Data collection tools included the Technostress Inventory by Tarafdar and colleagues (2007), Janssen's Individual Innovation Questionnaire (2000), and the Job Satisfaction Survey (JSS) by Spector (1994). The validity and reliability of these questionnaires were confirmed. The sample size for this study was 215, and data analysis was performed using SPSS and SMART-PLS software. Findings: Job satisfaction has a significant and positive relationship with individual innovation, idea generation, idea promotion, and idea implementation. Technostress moderates the relationship between job satisfaction and individual innovation, as well as idea generation and idea promotion. However, technostress does not play a moderating role in the relationship between job satisfaction and idea implementation. Conclusion: Based on the obtained results, organizations should take necessary measures to increase job satisfaction and reduce technostress among their employees. △ Less

Submitted 20 October, 2023; originally announced October 2023.

Comments: 13 pages, 2 figures, 4 tables

RD-Suite: A Benchmark for Ranking Distillation

Authors: Zhen Qin, Rolf Jagerman, Rama Pasumarthi, Honglei Zhuang, He Zhang, Aijun Bai, Kai Hui, Le Yan, Xuanhui Wang

Abstract: The distillation of ranking models has become an important topic in both academia and industry. In recent years, several advanced methods have been proposed to tackle this problem, often leveraging ranking information from teacher rankers that is absent in traditional classification settings. To date, there is no well-established consensus on how to evaluate this class of models. Moreover, inconsi… ▽ More The distillation of ranking models has become an important topic in both academia and industry. In recent years, several advanced methods have been proposed to tackle this problem, often leveraging ranking information from teacher rankers that is absent in traditional classification settings. To date, there is no well-established consensus on how to evaluate this class of models. Moreover, inconsistent benchmarking on a wide range of tasks and datasets make it difficult to assess or invigorate advances in this field. This paper first examines representative prior arts on ranking distillation, and raises three questions to be answered around methodology and reproducibility. To that end, we propose a systematic and unified benchmark, Ranking Distillation Suite (RD-Suite), which is a suite of tasks with 4 large real-world datasets, encompassing two major modalities (textual and numeric) and two applications (standard distillation and distillation transfer). RD-Suite consists of benchmark results that challenge some of the common wisdom in the field, and the release of datasets with teacher scores and evaluation scripts for future research. RD-Suite paves the way towards better understanding of ranking distillation, facilities more research in this direction, and presents new challenges. △ Less

Submitted 12 June, 2023; v1 submitted 7 June, 2023; originally announced June 2023.

Comments: 15 pages, 2 figures. arXiv admin note: text overlap with arXiv:2011.04006 by other authors

ACM Class: H.3.3

Regression Compatible Listwise Objectives for Calibrated Ranking with Binary Relevance

Authors: Aijun Bai, Rolf Jagerman, Zhen Qin, Le Yan, Pratyush Kar, Bing-Rong Lin, Xuanhui Wang, Michael Bendersky, Marc Najork

Abstract: As Learning-to-Rank (LTR) approaches primarily seek to improve ranking quality, their output scores are not scale-calibrated by design. This fundamentally limits LTR usage in score-sensitive applications. Though a simple multi-objective approach that combines a regression and a ranking objective can effectively learn scale-calibrated scores, we argue that the two objectives are not necessarily com… ▽ More As Learning-to-Rank (LTR) approaches primarily seek to improve ranking quality, their output scores are not scale-calibrated by design. This fundamentally limits LTR usage in score-sensitive applications. Though a simple multi-objective approach that combines a regression and a ranking objective can effectively learn scale-calibrated scores, we argue that the two objectives are not necessarily compatible, which makes the trade-off less ideal for either of them. In this paper, we propose a practical regression compatible ranking (RCR) approach that achieves a better trade-off, where the two ranking and regression components are proved to be mutually aligned. Although the same idea applies to ranking with both binary and graded relevance, we mainly focus on binary labels in this paper. We evaluate the proposed approach on several public LTR benchmarks and show that it consistently achieves either best or competitive result in terms of both regression and ranking metrics, and significantly improves the Pareto frontiers in the context of multi-objective optimization. Furthermore, we evaluated the proposed approach on YouTube Search and found that it not only improved the ranking quality of the production pCTR model, but also brought gains to the click prediction accuracy. The proposed approach has been successfully deployed in the YouTube production system. △ Less

Submitted 21 August, 2023; v1 submitted 2 November, 2022; originally announced November 2022.

Reducing Training Sample Memorization in GANs by Training with Memorization Rejection

Authors: Andrew Bai, Cho-Jui Hsieh, Wendy Kan, Hsuan-Tien Lin

Abstract: Generative adversarial network (GAN) continues to be a popular research direction due to its high generation quality. It is observed that many state-of-the-art GANs generate samples that are more similar to the training set than a holdout testing set from the same distribution, hinting some training samples are implicitly memorized in these models. This memorization behavior is unfavorable in many… ▽ More Generative adversarial network (GAN) continues to be a popular research direction due to its high generation quality. It is observed that many state-of-the-art GANs generate samples that are more similar to the training set than a holdout testing set from the same distribution, hinting some training samples are implicitly memorized in these models. This memorization behavior is unfavorable in many applications that demand the generated samples to be sufficiently distinct from known samples. Nevertheless, it is unclear whether it is possible to reduce memorization without compromising the generation quality. In this paper, we propose memorization rejection, a training scheme that rejects generated samples that are near-duplicates of training samples during training. Our scheme is simple, generic and can be directly applied to any GAN architecture. Experiments on multiple datasets and GAN models validate that memorization rejection effectively reduces training sample memorization, and in many cases does not sacrifice the generation quality. Code to reproduce the experiment results can be found at $\texttt{https://github.com/jybai/MRGAN}$. △ Less

Submitted 21 October, 2022; originally announced October 2022.

Concept Gradient: Concept-based Interpretation Without Linear Assumption

Authors: Andrew Bai, Chih-Kuan Yeh, Pradeep Ravikumar, Neil Y. C. Lin, Cho-Jui Hsieh

Abstract: Concept-based interpretations of black-box models are often more intuitive for humans to understand. The most widely adopted approach for concept-based interpretation is Concept Activation Vector (CAV). CAV relies on learning a linear relation between some latent representation of a given model and concepts. The linear separability is usually implicitly assumed but does not hold true in general. I… ▽ More Concept-based interpretations of black-box models are often more intuitive for humans to understand. The most widely adopted approach for concept-based interpretation is Concept Activation Vector (CAV). CAV relies on learning a linear relation between some latent representation of a given model and concepts. The linear separability is usually implicitly assumed but does not hold true in general. In this work, we started from the original intent of concept-based interpretation and proposed Concept Gradient (CG), extending concept-based interpretation beyond linear concept functions. We showed that for a general (potentially non-linear) concept, we can mathematically evaluate how a small change of concept affecting the model's prediction, which leads to an extension of gradient-based interpretation to the concept space. We demonstrated empirically that CG outperforms CAV in both toy examples and real world datasets. △ Less

Submitted 5 February, 2024; v1 submitted 31 August, 2022; originally announced August 2022.

Comments: 21 pages, 7 figures, published in ICLR 2023

Snowmass2021 Whitepaper: Muonium to antimuonium conversion

Authors: Ai-Yu Bai, Yu Chen, Yukai Chen, Rui-Rui Fan, Zhilong Hou, Han-Tao Jing, Hai-Bo Li, Yang Li, Han Miao, Huaxing Peng, Alexey A. Petrov, Ying-Peng Song, Jian Tang, Jing-Yu Tang, Nikolaos Vassilopoulos, Sampsa Vihonen, Chen Wu, Tian-Yu Xing, Yu Xu, Ye Yuan, Yao Zhang, Guang Zhao, Shi-Han Zhao, Luping Zhou

Abstract: The spontaneous muonium to antimuonium conversion is one of the interesting charged lepton flavor violation processes. It serves as a clear indication of new physics and plays an important role in constraining the parameter space beyond Standard Model. MACE is a proposed experiment to probe such a phenomenon and expected to enhance the sensitivity to the conversion probability by more than two ord… ▽ More The spontaneous muonium to antimuonium conversion is one of the interesting charged lepton flavor violation processes. It serves as a clear indication of new physics and plays an important role in constraining the parameter space beyond Standard Model. MACE is a proposed experiment to probe such a phenomenon and expected to enhance the sensitivity to the conversion probability by more than two orders of magnitude from the current best upper constraint obtained by the PSI experiment two decades ago. Recent developments in the theoretical and experimental aspects to search for such a rare process are summarized. △ Less

Submitted 21 March, 2022; originally announced March 2022.

Comments: 13 pages, 10 figures, 3 tables, contribution to Snowmass 2021

Million.js: A Fast Compiler-Augmented Virtual DOM for the Web

Authors: Aiden Bai

Abstract: Interactive web applications created with declarative JavaScript User Interface (UI) libraries have increasingly dominated the modern internet. However, existing libraries are primarily made for run-time execution, and rely on the user to load and render web applications. This led us to create Million.js, a fast compiler-augmented virtual Document Object Model (DOM) for the web. Million.js reduces… ▽ More Interactive web applications created with declarative JavaScript User Interface (UI) libraries have increasingly dominated the modern internet. However, existing libraries are primarily made for run-time execution, and rely on the user to load and render web applications. This led us to create Million.js, a fast compiler-augmented virtual Document Object Model (DOM) for the web. Million.js reduces load time and time-to-interactive by creating a compiler to compute interactive regions of a web application before the user visits the page. The virtual DOM run-time optimizes interactive content through compiler flags, compute batching, scheduling, and reactive data primitives to achieve optimal performance. When benchmarked against the most popular virtual DOM libraries, Million.js resulted in 133% to 300% faster rendering and 2347\% faster load. In a real-world web application with both comparative benchmarks and an informal user study, Million.js loaded 35.11% faster after migrating from React. The findings show that web applications have the potential to be orders of magnitude faster through JavaScript UI libraries that use Million.js. △ Less

Submitted 1 January, 2023; v1 submitted 16 February, 2022; originally announced February 2022.

Comments: 8 pages, 12 figures. Accepted to ACM SAC

Cavity-Enhanced Linear Dichroism in a van der Waals Antiferromagnet

Authors: Huiqin Zhang, Zhuoliang Ni, Aofeng Bai, Frank Peiris, Liang Wu, Deep Jariwala

Abstract: Optical birefringence is a fundamental optical property of crystals widely used for filtering and beam splitting of photons. Birefringent crystals concurrently possess the property of linear dichroism (LD) that allows asymmetric propagation or attenuation of light with two different polarizations. This property of LD has been widely studied from small molecules to polymers and crystals but has rar… ▽ More Optical birefringence is a fundamental optical property of crystals widely used for filtering and beam splitting of photons. Birefringent crystals concurrently possess the property of linear dichroism (LD) that allows asymmetric propagation or attenuation of light with two different polarizations. This property of LD has been widely studied from small molecules to polymers and crystals but has rarely been engineered per will. Here, we use the newly discovered spin-charge coupling in van der Waals antiferromagnetic (AFM) insulator FePS3 to induce large in-plane optical anisotropy and consequently LD. We report that the LD in this AFM insulator is tunable both spectrally and magnitude-wise as a function of cavity coupling. We demonstrate near-unity LD in the visible-near infrared range in cavity-coupled FePS3 crystals and derive its dispersion as a function of cavity length and FePS3 thickness. Our results hold wide implications for use of cavity tuned LD as a diagnostic probe for strongly correlated quantum materials as well as opens new opportunities for miniaturized, on-chip beam-splitters and tunable filters. △ Less

Submitted 9 February, 2022; originally announced February 2022.

Comments: 14 pages, 5 figures

Journal ref: Nature Photonics 16, 311-317 (2022)

Structure and magnetic properties of melilite-type compounds RE2Be2GeO7 (RE = Pr, Nd, Gd-Yb) with Rare-Earth ions on Shastry-Sutherland lattice

Authors: Malik Ashtar Yuming Bai, Longmeng Xu, Zongtang Wan, Zijun Wei, Yong Liu, Mohsin Ali Marwat, Zhaoming Tian

Abstract: Rare-earth (RE) based frustrated magnets as typical systems of combining strong spin-orbit coupling, geometric frustration and anisotropic exchange interactions, can give rise to diverse exotic magnetic ground states such as quantum spin liquid (QSL). The discovery of new RE-based frustrated materials is crucial for exploring the exotic magnetic phases. Herein, we report the synthesis, structure a… ▽ More Rare-earth (RE) based frustrated magnets as typical systems of combining strong spin-orbit coupling, geometric frustration and anisotropic exchange interactions, can give rise to diverse exotic magnetic ground states such as quantum spin liquid (QSL). The discovery of new RE-based frustrated materials is crucial for exploring the exotic magnetic phases. Herein, we report the synthesis, structure and magnetic properties of a family of melilite-type RE2Be2GeO7 (RE = Pr, Nd, Gd-Yb) compounds crystallized in a tetragonal structure, where magnetic RE3+ ions lay out on Shastry-Sutherland lattice (SSL) within ab-plane and are well separated by nonmagnetic GeBe2O7 polyhedrons along c-axis. Temperature-dependent susceptibilities and isothermal magnetization M(H) measurements reveal that most RE2Be2GeO7 compounds except RE=Tb show no magnetic ordering down to 2 K despite the dominant antiferromagnetic (AFM) interactions, where Tb2Be2GeO7 undergoes AFM transition with Neel temperature TN~ 2.5 K and field-induced spin flop behaviors (T< TN). In addition, the calculated magnetic entropy change from the isothermal M(H) curves reveal a viable magnetocaloric effect (MCE) for RE2Be2GeO7 (RE =Gd, Dy) in liquid helium temperature regimes, Gd2Be2GeO7 shows maximum Sm up to 54.8 J K-1 Kg-1 at H= 7 T and Dy2Be2GeO7 has largest value Sm=16.1 J K-1 kg-1 at H= 2 T in this family. More excitingly, rich diversity of RE ions in this family enables an archetype for exploring exotic quantum magnetic phenomena with large variability of spin located on SSL lattice. △ Less

Submitted 31 December, 2020; originally announced January 2021.

Comments: 19 pages 8 figures

Defining phylogenetic networks using ancestral profiles

Authors: Allan Bai, Peter Erdos, Charles Semple, Mike Steel

Abstract: Rooted phylogenetic networks provide a more complete representation of the ancestral relationship between species than phylogenetic trees when reticulate evolutionary processes are at play. One way to reconstruct a phylogenetic network is to consider its `ancestral profile' (the number of paths from each ancestral vertex to each leaf). In general, this information does not uniquely determine the u… ▽ More Rooted phylogenetic networks provide a more complete representation of the ancestral relationship between species than phylogenetic trees when reticulate evolutionary processes are at play. One way to reconstruct a phylogenetic network is to consider its `ancestral profile' (the number of paths from each ancestral vertex to each leaf). In general, this information does not uniquely determine the underlying phylogenetic network. A recent paper considered a new class of phylogenetic networks called `orchard networks' where this uniqueness was claimed to hold. Here we show that an additional restriction on the network, that of being `stack-free', is required in order for the original uniqueness claim to hold. On the other hand, if the additional stack-free restriction is lifted, we establish an alternative result; namely, there is uniqueness within the class of orchard networks up to the resolution of vertices of high in-degree. △ Less

Submitted 30 November, 2020; originally announced December 2020.

Comments: 18 pages, 4 figures. arXiv admin note: text overlap with arXiv:1901.04064

MSC Class: 05C85; 92D15

RoboCup 2D Soccer Simulation League: Evaluation Challenges

Authors: Mikhail Prokopenko, Peter Wang, Sebastian Marian, Aijun Bai, Xiao Li, Xiaoping Chen

Abstract: We summarise the results of RoboCup 2D Soccer Simulation League in 2016 (Leipzig), including the main competition and the evaluation round. The evaluation round held in Leipzig confirmed the strength of RoboCup-2015 champion (WrightEagle, i.e. WE2015) in the League, with only eventual finalists of 2016 competition capable of defeating WE2015. An extended, post-Leipzig, round-robin tournament which… ▽ More We summarise the results of RoboCup 2D Soccer Simulation League in 2016 (Leipzig), including the main competition and the evaluation round. The evaluation round held in Leipzig confirmed the strength of RoboCup-2015 champion (WrightEagle, i.e. WE2015) in the League, with only eventual finalists of 2016 competition capable of defeating WE2015. An extended, post-Leipzig, round-robin tournament which included the top 8 teams of 2016, as well as WE2015, with over 1000 games played for each pair, placed WE2015 third behind the champion team (Gliders2016) and the runner-up (HELIOS2016). This establishes WE2015 as a stable benchmark for the 2D Simulation League. We then contrast two ranking methods and suggest two options for future evaluation challenges. The first one, "The Champions Simulation League", is proposed to include 6 previous champions, directly competing against each other in a round-robin tournament, with the view to systematically trace the advancements in the League. The second proposal, "The Global Challenge", is aimed to increase the realism of the environmental conditions during the simulated games, by simulating specific features of different participating countries. △ Less

Submitted 14 June, 2017; originally announced June 2017.

Comments: 12 pages, RoboCup-2017, Nagoya, Japan, July 2017

Multi-Object Tracking and Identification over Sets

Authors: Aijun Bai

Abstract: The ability for an autonomous agent or robot to track and identify potentially multiple objects in a dynamic environment is essential for many applications, such as automated surveillance, traffic monitoring, human-robot interaction, etc. The main challenge is due to the noisy and incomplete perception including inevitable false negative and false positive errors from a low-level detector. In this… ▽ More The ability for an autonomous agent or robot to track and identify potentially multiple objects in a dynamic environment is essential for many applications, such as automated surveillance, traffic monitoring, human-robot interaction, etc. The main challenge is due to the noisy and incomplete perception including inevitable false negative and false positive errors from a low-level detector. In this paper, we propose a novel multi-object tracking and identification over sets approach to address this challenge. We define joint states and observations both as finite sets, and develop motion and observation functions accordingly. The object identification problem is then formulated and solved by using expectation-maximization methods. The set formulation enables us to avoid directly performing observation-to-object association. We empirically confirm that the overall algorithm outperforms the state-of-the-art in a popular PETS dataset. △ Less

Submitted 25 May, 2016; originally announced May 2016.

Comments: Draft version