Searcharxiv⌕ Search

arXiv subjects

Jiahong Li

Publications and source records attributed to Jiahong Li.

At least 19 recordsLinked to original sources

Hy ASR 3.0 Technical Report

We present Hy ASR 3.0, a large language model (LLM)-based automatic speech recognition (ASR) system that delivers strong recognition accuracy, supports flexible contextual conditioning, and maintains robust recognition performance across diverse speech conditions. Hy ASR 3.0 supports Mandarin Chinese, English and 13 Chinese dialect groups, including Cantonese. Among the evaluated systems, it achieves the lowest aggregate error rates on public Mandarin Chinese, English, and Cantonese benchmarks and the lowest error rates on the corresponding internal test sets, alongside first-tier dialect recognition performance. Beyond audio-only transcription, Hy ASR 3.0 supports multiple forms of external context, including hotword lists, question--answer dialogues, and summaries, enabling the use of task-relevant information during recognition. The model also leverages intra-utterance context to disambiguate homophonic and phonetically similar keywords using information from other parts of the same utterance, without requiring a separate contextual input. Evaluations demonstrate strong contextual recognition in both external-context and intra-utterance settings, as well as robust recognition of long-form speech, children's and older adults' speech, low-volume speech, and speech in noisy environments. This report describes the model architecture and training methodology and presents evaluations across recognition accuracy, context utilization, and robustness.

cs.CL↗

AzeroS: Extending LLM to Speech with Self-Generated Instruction-Free Tuning

Extending large language models (LLMs) to the speech domain has recently gained significant attention. A typical approach connects a pretrained LLM with an audio encoder through a projection module and trains the resulting model on large-scale, task-specific instruction-tuning datasets. However, curating such instruction-tuning data for specific requirements is time-consuming, and models trained in this manner often generalize poorly to unseen tasks. In this work, we first formulate that the strongest generalization of a speech-LLM is achieved when it is trained with Self-Generated Instruction-Free Tuning (SIFT), in which supervision signals are generated by a frozen LLM using textual representations of speech as input. Our proposed SIFT paradigm eliminates the need for collecting task-specific question-answer pairs and yields the theoretically best generalization to unseen tasks. Building upon this paradigm, we introduce AZeroS (Auden Zero-instruction-tuned Speech-LLM), which is trained on speech-text pairs derived from publicly available corpora, including approximately 25,000 hours of speech with ASR transcripts and 3,000 hours of speech with paralinguistic labels. Built upon Qwen2.5-7B-Instruct, the model updates only two lightweight projection modules (23.8 million parameters each), while keeping both the LLM and audio encoders frozen. Despite the minimal training cost and modest data scale, AZeroS achieves state-of-the-art performance on both semantic and paralinguistic benchmarks, including VoiceBench, AIR-Bench Foundation (Speech), and AIR-Bench Chat (Speech).

cs.CL↗

TaQFormer: A Treatment-as-Query Transformer with Data-Fusion for Multidimensional Treatment Modeling in Coupon Allocation

We study coupon allocation on an online marketplace offering multiple service categories. Issuing coupons requires choosing a vector of category-specific discounts, accounting for both their direct effects and cross-category demand shifts. We present TaQFormer, a treatment-as-query Transformer for this multidimensional treatment setting. Each treatment coordinate is represented as a query token, allowing cross-attention to estimate the response to the full treatment vector. To combine large but policy-biased observational logs with a smaller stream of randomized controlled trial data, we use a two-stage pipeline: observational pretraining learns general outcome patterns, followed by LoRA fine-tuning on RCT data with a request-specific fusion gate. We evaluate the resulting system through the checks required for production, including offline uplift metrics, treatment-parallel serving, latency monitoring, and a 14-day online A/B test. At essentially the same subsidy cost rate, \model increases business value per user by 1.26\% over the production baseline and significantly outperforms both the production baseline and the RCT-only model, while meeting the latency budget. It was subsequently rolled out to all marketplace traffic and has operated at full scale for several months as of October 2026. The deployment illustrates how multidimensional treatment modeling and observational pretraining, combined with adaptation using RCT data, can support effective joint discount decisions in production.

cs.LG↗

On the Nonoccurrence of Gradient Restart in Nesterov Accelerated Gradient Method

Gradient restarting often improves the practical performance of accelerated gradient methods for unconstrained convex optimization problems, but a general theoretical guarantee of such improvement remains unclear. In this paper, we construct smooth, strongly convex objectives with Lipschitz-continuous gradients and corresponding initial points for which gradient restart never occurs, both for Nesterov's accelerated gradient method and for its ordinary differential equation model. From these examples, we conclude that strong convexity and smoothness alone do not guarantee that gradient restart improves the convergence behavior of the accelerated method or its continuous-time model. In particular, for the continuous-time example, the objective gap is asymptotic to $1/(πt^3)$. We therefore obtain a negative answer to the open problem raised by Su, Boyd, and Cand{è}s [\textit{J. Mach. Learn. Res.}, 2016, 17(153), 1--43] of whether gradient restart guarantees linear convergence of the continuous dynamics for smooth strongly convex objectives.

math.OC↗

Continuous-wave 148.4-nm generation in top-seeded solution-grown strontium tetraborate

We report an implementation of continuous-wave (CW) vacuum-ultraviolet (VUV) generation at 148.4 nm by single-pass second-harmonic generation of 296.8 nm light in strontium tetraborate (SBO). At an incident fundamental power of approximately 580~mW, we estimate a VUV output power of approximately 1.7 nW at the crystal exit. The crystal is grown by the top-seeded solution growth method and operated in a helium-filled chamber. The quadratic power dependence observed under vacuum, together with its strong suppression in air, supports the identification of the generated second harmonic. We further characterize the dependence of the VUV signal on beam position, incidence angle, polarization, and crystal temperature, providing practical guidance for SBO-based CW VUV sources for nuclear spectroscopy and clock applications.

physics.optics↗

Oops, Not Now: PEARL, a RAG-Based Support Agent for Gameplay and What Players Want from AI Help

AI-powered gameplay support agents hold promise for game-based learning, yet grounding generative models in structured game data remains an open challenge. We present PEARL (Parallel Education Agent for Reflection and Learning), a dual-component Retrieval-Augmented Generation (RAG) system that combines semantic knowledge retrieval with structural board-state matching to deliver contextualized scaffolding in Parallel, a puzzle game for learning parallel programming. PEARL operates on two input streams (natural language queries and board topology), retrieving both conceptual explanations of gameplay moves and peer-generated board states as evidence: capabilities unavailable to a standard Large Language Model (LLM) with game state access alone. In a qualitative evaluation (N=10) comparing PEARL against an existing community-based Open Player Model (OPM) visualization system, participants preferred the visualization system on perceived usefulness and reported higher frustration with PEARL; five of ten minimized or abandoned the AI tool during play. Proactive delivery, generic responses, and trust deficits drove disengagement, while a subset of four participants found PEARL's grounded explanations complementary to visualization in specific contexts where they initiated the interaction. We position PEARL as a deployed design probe whose failure modes inform a concrete design agenda for AI gameplay support, captured as seven open problems for the community.

cs.HC↗

Alignment Is All You Need: Instruction-Free Training for General Audio-Language Models

Multimodal large language models (MLLMs) are typically built through a multi-stage pipeline consisting of cross-modal alignment, supervised fine-tuning (SFT), and preference optimization. This pipeline assumes that adapting an LLM to a new modality requires extensive task-specific supervision. However, pretrained LLMs already possess strong reasoning and instruction-following abilities. As LLMs evolve rapidly, an important question remains: can we efficiently transfer these capabilities to a new modality with minimal intervention, and is alignment alone sufficient for building a multimodal model? We introduce an Instruction-Free Alignment-Only large audio-language model (LALM) that keeps both the audio encoder and the LLM fully frozen, learning only a lightweight projector. Borrowing insights from AzeroS [1], we train on (audio, response) pairs from Self-Generated Data Construction, where an LLM expands captions into free-form responses without explicit task instructions. Across MMAU, MMAR, MMSU, and MMAU-Pro, our approach matches or surpasses heavily post-trained baselines using substantially less data. By keeping the LLM frozen, our model preserves its native instruction-following competence and can port seamlessly across model generations. Our results suggest that competitive MLLM can emerge from alignment alone, reducing multimodal extension to a lightweight projector-training problem that generalizes across modalities and adapts rapidly to each new LLM release.

cs.CL↗

Targeted Regularization for Causal Effect Estimation with Exponential Dispersion Family Outcomes

Neural Networks (NNs) for causal effect estimation have shown strong empirical performance, yet endowing them with desirable semiparametric properties -- doubly robustness and fast convergence rates -- remains challenging. A common approach to address this is targeted regularization, which modifies the objective function of NNs. However, existing work on neural causal effect estimation is largely limited to continuous outcomes, restricting its applicability to settings involving binary, count, or other skewed outcomes commonly encountered in practice. We propose a unified targeted regularization framework for the Exponential Dispersion Family (EDF) to address this limitation. Specifically, we first derive the von Mises expansion of the average dose function of canonical functions (ADCF) for discrete treatments and of the sieve-projected ADCF for continuous treatments. Second, we use this expansion to construct a unified targeted regularization, that corrects first-order bias at the distributional level. We integrate this objective into a NN architecture that jointly estimates the outcome model, propensity score model, and fluctuation parameter end-to-end. Experimental results demonstrate the effectiveness of our method.

cs.LG↗

A Brief Overview: On-Policy Self-Distillation In Large Language Models

On-Policy Self-Distillation (OPSD) is a unified learning framework in which a single large language model acts simultaneously as both teacher and student. Unlike conventional knowledge distillation that relies on a separate, often larger teacher model, OPSD operates under different contextual roles: the teacher policy is granted privileged access to verified reasoning traces, while the student policy observes only the problem statement. OPSD is trained to minimize per-token distributional divergence between the two roles over trajectories sampled from the student itself, thereby aligning its own reasoning behavior with solution-aware rationalizations. OPSD eliminates the need for an external teacher, directly leverages ground-truth solution information, and resolves the distribution mismatch inherent in off-policy distillation. OPSD typically reduces GPU memory consumption by approximately 40%-60% compared to standard On-Policy Distillation (OPD). In this paper, we present a brief analysis of the conceptual foundations, methodological innovations, and principled designs underlying recent advances in OPSD for large language models. This discussion, crafted from the perspective of beginners in this field, aims to provide a concise overview of the design principles and emerging patterns of OPSD in LLMs, intended for researchers who are similarly new to this area.

cs.HC↗

Rethinking Agentic Reinforcement Learning In Large Language Models

Reinforcement Learning (RL) has traditionally focused on training specialized agents to optimize predefined reward functions within narrowly defined environments. However, the advent of powerful Large Language Models (LLMs) and increasingly complex, open-ended tasks has catalyzed a paradigm shift towards agentic paradigms within RL. This emerging framework extends beyond traditional RL by emphasizing the development of autonomous agents capable of goal-setting, long-term planning, dynamic strategy adaptation, and interactive reasoning in uncertain, real-world environments. Unlike conventional approaches that rely heavily on static objectives and episodic interactions, LLM-based Agentic RL incorporates cognitive-like capabilities such as meta-reasoning, self-reflection, and multi-step decision-making directly into the learning loop. In this paper, we provide a deep insight for looking the conceptual foundations, methodological innovations, and effective designs underlying this trend. Furthermore, we identify critical challenges and outline promising future directions for building LLM-based Agentic RL.

cs.AI↗

TransSplat: Unbalanced Semantic Transport for Language-Driven 3DGS Editing

Language-driven 3D Gaussian Splatting (3DGS) editing provides a more convenient approach for modifying complex scenes in VR/AR. Standard pipelines typically adopt a two-stage strategy: first editing multiple 2D views, and then optimizing the 3D representation to match these edited observations. Existing methods mainly improve view consistency through multi-view feature fusion, attention filtering, or iterative recalibration. However, they fail to explicitly address a more fundamental issue: the semantic correspondence between edited 2D evidence and 3D Gaussians. To tackle this problem, we propose TransSplat, which formulates language-driven 3DGS editing as a multi-view unbalanced semantic transport problem. Specifically, our method establishes correspondences between visible Gaussians and view-specific editing prototypes, thereby explicitly characterizing the semantic relationship between edited 2D evidence and 3D Gaussians. It further recovers a cross-view shared canonical 3D edit field to guide unified 3D appearance updates. In addition, we use transport residuals to suppress erroneous edits in non-target regions, mitigating edit leakage and improving local control precision. Qualitative and quantitative results show that, compared with existing 3D editing methods centered on enhancing view consistency, TransSplat achieves superior performance in local editing accuracy and structural consistency.

cs.CV↗

Unlocking Open-Player-Modeling-enhanced Game-Based Learning: The Open Player Socially Analytical Intelligence Architecture

Game-Based Learning (GBL) is a learner-engaging pedagogical methodology, yet adapting games to heterogeneous learners requires transparent, real-time Open Player Models (OPMs). We contribute to the community Open Player Socially Analytical Intelligence (OPSAI), an architecture implementing OPM beyond conceptual frameworks and validated in a GBL application. It decouples gameplay telemetry and analysis from the game engine and automatically derives pedagogically actionable insights, supporting the transparency of computational player models while making them accessible to players. OPSAI comprises three logical layers: a Frontend that both provides the GBL experience and collects information needed for analytics; a stateless Backend that hosts transparent analytics services producing reflective prompts, recommendations, and visualization guides; and a two-tier Log Storage that balances heavy raw gameplay data with lightweight reference indices for low-latency queries. By feeding analytics outputs back into the game interface, OPSAI closes the feedback loop between play and learning, empowering teachers, researchers, and learners alike. We further showcase OPSAI with a full deployment on the Parallel GBL environment, featuring live play traces, peer comparisons, and personalized suggestions, demonstrating a reusable blueprint for future educational games.

cs.HC↗

Outlier-robust Autocovariance Least Square Estimation via Iteratively Reweighted Least Square

The autocovariance least squares (ALS) method is a computationally efficient approach for estimating noise covariances in Kalman filters without requiring specific noise models. However, conventional ALS and its variants rely on the classic least mean squares (LMS) criterion, making them highly sensitive to measurement outliers and prone to severe performance degradation. To overcome this limitation, this paper proposes a novel outlier-robust ALS algorithm, termed ALS-IRLS, based on the iteratively reweighted least squares (IRLS) framework. Specifically, the proposed approach introduces a two-tier robustification strategy. First, an innovation-level adaptive thresholding mechanism is employed to filter out heavily contaminated data. Second, the outlier-contaminated autocovariance is formulated using an $ε$-contamination model, where the standard LMS criterion is replaced by the Huber cost function. The IRLS method is then utilized to iteratively adjust data weights based on estimation deviations, effectively mitigating the influence of residual outliers. Comparative simulations demonstrate that ALS-IRLS reduces the root-mean-square error (RMSE) of noise covariance estimates by over two orders of magnitude compared to standard ALS. Furthermore, it significantly enhances downstream state estimation accuracy, outperforming existing outlier-robust Kalman filters and achieving performance nearly equivalent to the ideal Oracle lower bound in the presence of noisy and anomalous data.

math.OC↗

TTA: Transcribe, Translate and Alignment for Cross-lingual Speech Representation

Speech-LLM models have demonstrated great performance in multi-modal and multi-task speech understanding. A typical speech-LLM paradigm is integrating speech modality with a large language model (LLM). While the Whisper encoder was frequently adopted in previous studies for speech input, it shows limitations regarding input format, model scale, and semantic performance. To this end, we propose a lightweight TTA model specialized in speech semantics for more effective LLM integration. With large-scale training of 358k hours of speech data on multilingual speech recognition (ASR), speech translation (ST) and speech-text alignment tasks, TTA is capable of producing robust cross-lingual speech representations. Extensive evaluations across diverse benchmarks, including ASR/ST, speech retrieval, and ASR-LLM performance assessments, demonstrate TTA's superiority over Whisper. Furthermore, we rigorously validate the interplay between cross-lingual capabilities and ASR/ST performance. The model weights and training recipes of TTA will be released as part of an audio understanding toolkit Auden.

eess.AS↗

Towards Comprehensive Semantic Speech Embeddings for Chinese Dialects

Despite having hundreds of millions of speakers, Chinese dialects lag behind Mandarin in speech and language technologies. Most varieties are primarily spoken, making dialect-to-Mandarin speech-LLMs (large language models) more practical than dialect LLMs. Building dialect-to-Mandarin speech-LLMs requires speech representations with cross-dialect semantic alignment between Chinese dialects and Mandarin. In this paper, we achieve such a cross-dialect semantic alignment by training a speech encoder with ASR (automatic speech recognition)-only data, as demonstrated by speech-to-speech retrieval on a new benchmark of spoken Chinese varieties that we contribute. Our speech encoder further demonstrates state-of-the-art ASR performance on Chinese dialects. Together, our Chinese dialect benchmark, semantically aligned speech representations, and speech-to-speech retrieval evaluation lay the groundwork for future Chinese dialect speech-LLMs. We release the benchmark at https://github.com/kalvinchang/yubao.

cs.CL↗

Efficient Multilingual ASR Finetuning via LoRA Language Experts

Recent advancements in deep learning have significantly enhanced multilingual automatic speech recognition (ASR) due to the development of advanced model architectures and available large-scale multilingual datasets. Despite that, multilingual ASR still suffers from the curse of multilinguality in that different languages tend to interfere with each other, making it difficult for the ASR model to identify multiple languages effectively while sharing model capacity across them. This paper proposes an efficient finetuning framework for customized multilingual ASR via prepared LoRA language experts based on Whisper. Through LoRA expert fusion or knowledge distillation, our approach achieves better recognition performance on target languages than standard fine-tuning methods. Experimental results demonstrate that the proposed models yield approximately 10\% and 15\% relative performance gains in language-aware and language-agnostic scenarios, respectively.

cs.CL↗

The Global R-linear Convergence of Nesterov's Accelerated Gradient Method with Unknown Strongly Convex Parameter

The Nesterov accelerated gradient (NAG) method is an important extrapolation-based numerical algorithm that accelerates the convergence of the gradient descent method in convex optimization. When dealing with an objective function that is $μ$-strongly convex, selecting extrapolation coefficients dependent on $μ$ enables global R-linear convergence. In cases where $μ$ is unknown, a commonly adopted approach is to set the extrapolation coefficient using the original NAG method. This choice allows for achieving the optimal iteration complexity among first-order methods for general convex problems. However, it remains unknown whether the NAG method with an unknown strongly convex parameter exhibits global R-linear convergence for strongly convex problems. In this work, we answer this question positively by establishing the Q-linear convergence of certain constructed Lyapunov sequences. Furthermore, we extend our result to the global R-linear convergence of the accelerated proximal gradient method, which is employed for solving strongly convex composite optimization problems. Interestingly, these results contradict the findings of the continuous counterpart of the NAG method in [Su, Boyd, and Candés, J. Mach. Learn. Res., 2016, 17(153), 1-43], where the convergence rate by the suggested ordinary differential equation cannot exceed the $O(1/{\tt poly}(k))$ for strongly convex functions.

math.OC↗

Accelerated Gradient Methods with Gradient Restart: Global Linear Convergence

Gradient restarting has been shown to improve the numerical performance of accelerated gradient methods. This paper provides a mathematical analysis to understand these advantages. First, we establish global linear convergence guarantees for both the original and gradient restarted accelerated proximal gradient method when solving strongly convex composite optimization problems. Second, through analysis of the corresponding ordinary differential equation model, we prove the continuous trajectory of the gradient restarted Nesterov's accelerated gradient method exhibits global linear convergence for quadratic convex objectives, while the non-restarted version provably lacks this property by [Su, Boyd, and Candés, \textit{J. Mach. Learn. Res.}, 2016, 17(153), 1-43].

math.OC↗