arXiv subjects

Fei Huang

Publications and source records attributed to Fei Huang.

At least 127 records · Page 7Linked to original sources

arXiv2025

Qwen3Guard Technical Report

As large language models (LLMs) become more capable and widely used, ensuring the safety of their outputs is increasingly critical. Existing guardrail models, though useful in static evaluation settings, face two major limitations in real-world applications: (1) they typically output only binary "safe/unsafe" labels, which can be interpreted inconsistently across diverse safety policies, rendering them incapable of accommodating varying safety tolerances across domains; and (2) they require complete model outputs before performing safety checks, making them fundamentally incompatible with streaming LLM inference, thereby preventing timely intervention during generation and increasing exposure to harmful partial outputs. To address these challenges, we present Qwen3Guard, a series of multilingual safety guardrail models with two specialized variants: Generative Qwen3Guard, which casts safety classification as an instruction-following task to enable fine-grained tri-class judgments (safe, controversial, unsafe); and Stream Qwen3Guard, which introduces a token-level classification head for real-time safety monitoring during incremental text generation. Both variants are available in three sizes (0.6B, 4B, and 8B parameters) and support up to 119 languages and dialects, providing comprehensive, scalable, and low-latency safety moderation for global LLM deployments. Evaluated across English, Chinese, and multilingual benchmarks, Qwen3Guard achieves state-of-the-art performance in both prompt and response safety classification. All models are released under the Apache 2.0 license for public use.

cs.CL↗

arXiv2025

Rethinking LLM Evaluation: Can We Evaluate LLMs with 200x Less Data?

As the demand for comprehensive evaluations of diverse model capabilities steadily increases, benchmark suites have correspondingly grown significantly in scale. Despite notable advances in redundancy reduction and subset-level performance prediction, a systematic framework that effectively integrates these methods to ensure both prediction accuracy and ranking consistency is still largely elusive. In this paper, we first perform a sample-level analysis of benchmark redundancy and identify several highly similar samples that can be eliminated. Besides, we frame benchmark compression as an optimization problem with the aim of score reconstruction. Building on these, we then propose EssenceBench, a coarse-to-fine framework utilizing an iterative Genetic Algorithm (GA), which takes the advantages of fitness-based subset search and attribution-based sample search. Compared to previous methods, our approach yields superior compression results with lower reconstruction error and markedly higher efficiency. In particular, on the HellaSwag benchmark (10K samples), our method preserves the ranking of all models shifting within 5% using 25x fewer samples, and achieves 95% ranking preservation shifting within 5% using only 200x fewer samples.

cs.CL↗

arXiv2025

Nature of the $P_c$ states from compositeness criteria

Based on a coupled-channel approach, we investigate the structures of four $P_c$ states through compositeness criteria. Toward a more precise description of the states, we have obtained refined fit results of the LHCb data on the $J/ψp$ invariant mass distribution of the $Λ_b^0\to J/ψp K^-$ decay. Allowing for the fact that each of the four $P_c$ states couples strongly to a nearby $S$-wave channel, three criteria on the compositeness/elementariness are adopted in this study: the pole-counting rule, the spectral density function, and the Gamow wave function. Compositeness information is extracted from the scattering amplitudes and the pole parameters (pole positions and residues), without any preconceived assumptions on the nature of the $P_c$ states, and without any dependence on the model parametrization. Consistently within the framework of all the three methods, it has been found that the $P_c(4312)\,1/2^-$ is mainly composed by $\bar{D}Σ_c$, $P_c(4380)\,3/2^-$ by $\bar{D}Σ_c^*$, while the $P_c(4440)\,1/2^-$ and $P_c(4457)\,3/2^-$ states both turn out as composite states of $\bar{D}^*Σ_c$. The upper limits of the values of their elementariness are estimated to be rather small. This paper provides an additional confirmation of the molecular interpretation for the $P_c$ states in the literature.

hep-ph↗

arXiv2025

Primordial Black Holes Place the Universe in Stasis

A variety of scenarios for early-universe cosmology give rise to a population of primordial black holes (PBHs) with a broad spectrum of masses. The evaporation of PBHs in such scenarios has the potential to place the universe into an extended period of "stasis" during which the abundances of matter and radiation remain absolutely constant despite cosmological expansion. This surprising phenomenon can give rise to new possibilities for early-universe dynamics and lead to distinctive signatures of the evaporation of such PBHs. In this paper, we discuss how this stasis epoch arises and explore a number of its phenomenological consequences, including implications for inflationary observables, the stochastic gravitational-wave background, baryogenesis, and the production of dark matter and dark radiation.

astro-ph.CO↗

arXiv2025

New Physics Search at the CEPC: a General Perspective

The Circular Electron-Positron Collider (CEPC), a proposed next-generation Higgs factory, provides new opportunities to explore physics beyond the Standard Model (SM). With its clean electron-positron collision environment and the ability to collect large samples of Higgs, W, and Z bosons, the CEPC enables precision measurements and searches for new physics. This white paper outlines the CEPC's discovery potential, including studies of exotic decays of the Higgs, Z, and top quarks, dark matter and dark sector phenomena, long-lived particles, supersymmetry, and neutrino-related signatures. Advanced detector technologies and reconstruction techniques, such as one-to-one correspondence reconstruction and jet origin identification, significantly improve sensitivity to rare and weakly interacting processes. The CEPC is particularly well suited to probe the electroweak phase transition and test models of electroweak baryogenesis and dark sector interactions. In addition, global fit analyses highlight the CEPC's complementary role in constraining a wide range of new physics scenarios. These features position the CEPC as a powerful tool for exploring the next frontier in fundamental particle physics in the post-Higgs discovery era.

hep-ex↗

arXiv2025

PARL-MT: Learning to Call Functions in Multi-Turn Conversation with Progress Awareness

Large language models (LLMs) have achieved impressive success in single-turn function calling, yet real-world applications such as travel planning or multi-stage data analysis typically unfold across multi-turn conversations. In these settings, LLMs must not only issue accurate function calls at each step but also maintain progress awareness, the ability to summarize past interactions and plan future actions to ensure coherent, long-horizon task execution. Existing approaches, however, either reduce multi-turn training to isolated single-turn samples, which neglects task-level planning, or employ end-to-end reinforcement learning (RL) that struggles with redundancy and lacks explicit integration of progress awareness. To overcome these limitations, we introduce PARL-MT, a framework that explicitly incorporates progress awareness into LLM training for multi-turn function calling. PARL-MT combines (i) a Progress Awareness Generation (PAG) pipeline, which automatically constructs datasets coupling conversation summaries with future task planning, and (ii) a Progress Awareness-Guided Reinforcement Learning (PAG-RL) algorithm, which integrates progress awareness into RL training to reduce contextual redundancy and improve alignment between local actions and global task completion. Empirical results on two public benchmarks demonstrate that PARL-MT significantly outperforms existing methods, highlighting the effectiveness of progress awareness in enabling robust and efficient multi-turn function calling.

cs.CL↗

arXiv2025

Primordial Black Holes and their Mass Spectra: The Effects of Mergers and Accretion within Stasis Cosmologies

A variety of processes in the very early universe can give rise to a population of primordial black holes (PBHs) with an extended mass spectrum. For certain mass spectra of this sort, it has been shown that the evaporation of these PBHs into radiation can drive the universe toward an epoch of cosmological stasis which can persist for a significant number of $e$-folds of cosmological expansion. However, in general, the initial mass spectrum which characterizes a population of PBHs at the time of production can subsequently be distorted by processes such as mergers and accretion. In this paper, we examine the effects that these processes have on the spectra that lead to a PBH-induced stasis. Within such stasis models, we find that mergers have only a negligible effect on these spectra within the regime of interest for stasis. We likewise find that the effect of accretion is negligible in many cases of interest. However, we find that the effect of accretion on the PBH mass spectrum is non-negligible in situations in which this spectrum is particularly broad. In such situations, the stasis epoch is abridged or, in extreme cases, does not occur at all. Thus accretion plays a non-trivial role in constraining the emergence of stasis within scenarios which lead to extended PBH mass spectra.

astro-ph.CO↗

arXiv2025

WebWeaver: Structuring Web-Scale Evidence with Dynamic Outlines for Open-Ended Deep Research

This paper tackles \textbf{open-ended deep research (OEDR)}, a complex challenge where AI agents must synthesize vast web-scale information into insightful reports. Current approaches are plagued by dual-fold limitations: static research pipelines that decouple planning from evidence acquisition and monolithic generation paradigms that include redundant, irrelevant evidence, suffering from hallucination issues and low citation accuracy. To address these challenges, we introduce \textbf{WebWeaver}, a novel dual-agent framework that emulates the human research process. The planner operates in a dynamic cycle, iteratively interleaving evidence acquisition with outline optimization to produce a comprehensive, citation-grounded outline linking to a memory bank of evidence. The writer then executes a hierarchical retrieval and writing process, composing the report section by section. By performing targeted retrieval of only the necessary evidence from the memory bank via citations for each part, it effectively mitigates long-context issues and citation hallucinations. Our framework establishes a new state-of-the-art across major OEDR benchmarks, including DeepResearch Bench, DeepConsult, and DeepResearchGym. These results validate our human-centric, iterative methodology, demonstrating that adaptive planning and focused synthesis are crucial for producing comprehensive, trusted, and well-structured reports.

cs.CL↗

arXiv2025

Effects of hadronic molecule $N(2080)3/2^-$ on $K^{*+}Λ$ photoproduction

In our previous work [Phys. Rev. C {\bf 101}, 014003 (2020)], we have analyzed all available data on differential cross sections and spin density matrix elements for the $γp \to K^{\ast +} Λ$ reaction using an effective Lagrangian approach. There, the $t$-channel $K$, $K^*$, and $κ$ exchanges, the $u$-channel $Λ$, $Σ$, and $Σ^*$ exchanges, the $s$-channel $N$, $N(2060)5/2^-$, and $N(2000)5/2^+$ exchanges, and the interaction current were taken into account in constructing the reaction amplitudes. It was found that, overall, the agreement of the theoretical results with the corresponding data is fairly good. However, noticeable discrepancies between the model results and the data on spin density matrix elements $ρ_{00}$ are still observed in the center-of-mass energy region around $W \approx 2.08$ GeV, indicating the need for additional reaction mechanisms. On the other hand, the hidden-strange pentaquark-like state $N(2080)3/2^-$, which is proposed to be a $K^*Σ$ molecular state as the strange partner of $P_c(4457)$ hadronic molecule, has been found to play quite active roles in reproducing the data for $K^{*+}Σ^0$, $K^{*0}Σ^+$, and $ϕp$ photoproduction reactions. Based on these observations, in the present work, we reanalyze the data for the $γp \to K^{\ast +} Λ$ reaction by incorporating the pentaquark-like state $N(2080)3/2^-$ in our previously proposed model. The results show that with the inclusion of the contributions of $N(2080)3/2^-$, the quality of the theoretical description of the $γp \to K^{*+}Λ$ data can be considerably improved.

hep-ph↗

arXiv2025

Lepton Number Violation Higgs Decay at Muon Collider

In this paper, we consider the scalar singlet extension of type-I seesaw, where a scalar singlet $S$ and heavy neutral lepton $N$ are further introduced. The Majorana mass term of heavy neutral lepton is generated through the Yukawa interaction with the scalar singlet, which then induces the lepton number violation decays of SM Higgs $h$ and heavy Higgs $H$ via mixing of scalars. As a pathway to probe the origin of heavy neutral lepton mass, we investigate the lepton number violation Higgs decay signature at the TeV-scale muon collider. The dominant production channel of Higgs bosons at the TeV-scale muon collider is via vector boson fusion. So we perform a detailed analysis of the signal process $μ^+μ^-\to ν_μ\barν_μh/H \to ν_μ\barν_μNN $ followed by $ N \to μ^\pm jj$, where the two jets from $W$ boson decay are treated as one fat-jet $J$. With an integrated luminosity of $1(10)~\text{ab}^{-1}$, the 3 (10) TeV muon collider could discover the lepton number violation SM Higgs decay $h\to μ^\pmμ^\pm JJ$ signature for the Higgs mixing parameter $\sinα>0.05(0.009)$. Meanwhile, a large parameter space can be detected by the lepton number violation heavy Higgs decay $H\to μ^\pmμ^\pm JJ$ signature for $m_H\lesssim1 (3)$ TeV and $\sinα\gtrsim0.03(0.005)$ at the 3 (10) TeV muon collider. Therefore, the lepton number violation SM and heavy Higgs decay signatures are both promising at the TeV scale muon collider.

hep-ph↗

arXiv2025

Agentic Reinforcement Learning with Implicit Step Rewards

Large language models (LLMs) are increasingly developed as autonomous agents using reinforcement learning (agentic RL) that reason and act in interactive environments. However, sparse and sometimes unverifiable rewards make it extremely challenging to assign credit when training LLM agents that serve as a policy. Recent work attempts to integrate process supervision into RL but suffers from biased annotation, reward hacking, high-variance from overly fine-grained rewards or failtures when state overlap is rare. We therefore introduce implicit step rewards for agentic RL (iStar), a general credit-assignment strategy that integrates seamlessly with standard RL algorithms without relying on additional rollouts or explicit step labels. Particularly, we alternatively optimize an implicit process reward model (PRM) with the policy model to generate implicit step rewards via a trajectory-based DPO objective. Theoretical analysis shows that this learning objective produces a step-wise reward function. Then the implicit step rewards are used to compute step-level advantages, which are combined with trajectory (or episode)-level advantages for policy updates, creating a self-reinforcing training loop. We evaluate our method on three challenging agent benchmarks, including WebShop and VisualSokoban, as well as open-ended social interactions with unverifiable rewards in SOTOPIA. Crucially, iStar shows superior performance over frontier LLMs and strong RL baselines across domains, achieving state-of-the-art results with higher sample-efficiency and training stability. Further analysis also demonstrates efficient exploration by iStar with increased rewards in both step- and episode-level while maintaining fewer steps to achieve task success. Code will be available soon.

cs.CL↗

arXiv2025

GSID: Generative Semantic Indexing for E-Commerce Product Understanding

Structured representation of product information is a major bottleneck for the efficiency of e-commerce platforms, especially in second-hand ecommerce platforms. Currently, most product information are organized based on manually curated product categories and attributes, which often fail to adequately cover long-tail products and do not align well with buyer preference. To address these problems, we propose \textbf{G}enerative \textbf{S}emantic \textbf{I}n\textbf{D}exings (GSID), a data-driven approach to generate product structured representations. GSID consists of two key components: (1) Pre-training on unstructured product metadata to learn in-domain semantic embeddings, and (2) Generating more effective semantic codes tailored for downstream product-centric applications. Extensive experiments are conducted to validate the effectiveness of GSID, and it has been successfully deployed on the real-world e-commerce platform, achieving promising results on product understanding and other downstream tasks.

cs.IR↗

arXiv2025

UI-S1: Advancing GUI Automation via Semi-online Reinforcement Learning

Graphical User Interface (GUI) agents have demonstrated remarkable progress in automating complex user interface interactions through reinforcement learning. However, current approaches face a fundamental dilemma: offline RL enables stable training on pre-collected trajectories, but struggles with multi-step task execution for lack of trajectory-level reward signals; online RL captures these signals through environment interaction, but suffers from sparse rewards and prohibitive deployment costs. To address it, we present Semi-online Reinforcement Learning, a novel paradigm that simulates online RL on offline trajectories. During each rollout process, we preserve the original model output within the multi-turn dialogue, where a Patch Module adaptively recovers the divergence between rollout and expert trajectories. To capture long-term training signals, Semi-online RL introduces discounted future returns into the reward computation and optimizes the policy with weighted step-level and episode-level advantages. We further introduce Semi-Online Performance (SOP), a metric that aligns better with true online performance, serving as a practical and effective proxy for real-world evaluation. Experiments show that ours Semi-online RL achieves SOTA performance among 7B models across four dynamic benchmarks, with significant gains over the base model (e.g., +12.0% on AndroidWorld, +23.8% on AITW), demonstrating significant progress in bridging the gap between offline training efficiency and online multi-turn reasoning. The code is available at https://github.com/X-PLUG/MobileAgent/tree/main/UI-S1.

cs.LG↗

arXiv2025

PART: Progressive Alignment Representation Training for Multilingual Speech-To-Text with LLMs

Large language models (LLMs) have expanded from text to speech, giving rise to Speech Large Models (SLMs) that support recognition, translation, and synthesis. A key challenge is aligning speech and text representations, which becomes harder in multilingual settings. Existing methods often freeze LLM parameters and train encoders on multilingual data, but this forces cross-language convergence and limits performance. We introduce Progressive Alignment Representation Training (PART), a multi-stage and multi-task framework that separates within-language from cross-language alignment. During cross-language training, LLM parameters are dynamically activated, and text-based tasks are later introduced to enhance multilingual understanding. Experiments on CommonVoice 15, Fleurs, Wenetspeech, and CoVoST2 show that PART surpasses conventional approaches, with analysis confirming its ability to balance language-specific distinctions and cross-language generalization. These results demonstrate PART's effectiveness and generality for multilingual speech modality alignment.

cs.CL↗

arXiv2025

OpenOmni: Advancing Open-Source Omnimodal Large Language Models with Progressive Multimodal Alignment and Real-Time Self-Aware Emotional Speech Synthesis

Recent advancements in omnimodal learning have significantly improved understanding and generation across images, text, and speech, yet these developments remain predominantly confined to proprietary models. The lack of high-quality omnimodal datasets and the challenges of real-time emotional speech synthesis have notably hindered progress in open-source research. To address these limitations, we introduce \name, a two-stage training framework that integrates omnimodal alignment and speech generation to develop a state-of-the-art omnimodal large language model. In the alignment phase, a pre-trained speech model undergoes further training on text-image tasks, enabling (near) zero-shot generalization from vision to speech, outperforming models trained on tri-modal datasets. In the speech generation phase, a lightweight decoder is trained on speech tasks with direct preference optimization, enabling real-time emotional speech synthesis with high fidelity. Experiments show that \name surpasses state-of-the-art models across omnimodal, vision-language, and speech-language benchmarks. It achieves a 4-point absolute improvement on OmniBench over the leading open-source model VITA, despite using 5x fewer training samples and a smaller model size (7B vs. 7x8B). Additionally, \name achieves real-time speech generation with <1s latency at non-autoregressive mode, reducing inference time by 5x compared to autoregressive methods, and improves emotion classification accuracy by 7.7\%

cs.CL↗

arXiv2025

WebResearcher: Unleashing unbounded reasoning capability in Long-Horizon Agents

Recent advances in deep-research systems have demonstrated the potential for AI agents to autonomously discover and synthesize knowledge from external sources. In this paper, we introduce WebResearcher, a novel framework for building such agents through two key components: (1) WebResearcher, an iterative deep-research paradigm that reformulates deep research as a Markov Decision Process, where agents periodically consolidate findings into evolving reports while maintaining focused workspaces, overcoming the context suffocation and noise contamination that plague existing mono-contextual approaches; and (2) WebFrontier, a scalable data synthesis engine that generates high-quality training data through tool-augmented complexity escalation, enabling systematic creation of research tasks that bridge the gap between passive knowledge recall and active knowledge construction. Notably, we find that the training data from our paradigm significantly enhances tool-use capabilities even for traditional mono-contextual methods. Furthermore, our paradigm naturally scales through parallel thinking, enabling concurrent multi-agent exploration for more comprehensive conclusions. Extensive experiments across 6 challenging benchmarks demonstrate that WebResearcher achieves state-of-the-art performance, even surpassing frontier proprietary systems.

cs.CL↗

arXiv2025

Enhancing LLM Language Adaption through Cross-lingual In-Context Pre-training

Large language models (LLMs) exhibit remarkable multilingual capabilities despite English-dominated pre-training, attributed to cross-lingual mechanisms during pre-training. Existing methods for enhancing cross-lingual transfer remain constrained by parallel resources, suffering from limited linguistic and domain coverage. We propose Cross-lingual In-context Pre-training (CrossIC-PT), a simple and scalable approach that enhances cross-lingual transfer by leveraging semantically related bilingual texts via simple next-word prediction. We construct CrossIC-PT samples by interleaving semantic-related bilingual Wikipedia documents into a single context window. To access window size constraints, we implement a systematic segmentation policy to split long bilingual document pairs into chunks while adjusting the sliding window mechanism to preserve contextual coherence. We further extend data availability through a semantic retrieval framework to construct CrossIC-PT samples from web-crawled corpus. Experimental results demonstrate that CrossIC-PT improves multilingual performance on three models (Llama-3.1-8B, Qwen2.5-7B, and Qwen2.5-1.5B) across six target languages, yielding performance gains of 3.79%, 3.99%, and 1.95%, respectively, with additional improvements after data augmentation.

cs.CL↗

arXiv2025

Translationese-index: Using Likelihood Ratios for Graded and Generalizable Measurement of Translationese

Translationese refers to linguistic properties that usually occur in translated texts. Previous works study translationese by framing it as a binary classification between original texts and translated texts. In this paper, we argue that translationese should be graded instead of binary and propose the first measure for translationese -- the translationese-index (T-index), computed from the likelihood ratios of two contrastively fine-tuned language models (LMs). We use synthesized translations and translations in the wild to evaluate T-index's generalizability in cross-domain settings and its validity against human judgments. Our results show that T-index can generalize to unseen genres, authors, and language pairs. Moreover, T-index computed using two 0.5B LMs fine-tuned on only 1-5k pairs of synthetic data can effectively capture translationese, as demonstrated by alignment with human pointwise ratings and pairwise judgments. Additionally, the correlation between T-index and existing machine translation (MT) quality estimation (QE) metrics such as BLEU and COMET is low, suggesting that T-index is not covered by these metrics and can serve as a complementary metric in MT QE.

cs.CL↗