arXiv subjects

Yunke Zhao

Publications and source records attributed to Yunke Zhao.

6 recordsLinked to original sources

arXiv2026

Cosmological Collider Signals at Strong Mixing

We study cosmological collider signatures in a two-field inflationary system with constant-turn derivative mixing between the canonically normalized curvature fluctuation and a massive isocurvature field. Building on recently derived exact hypergeometric solutions that treat the quadratic mixing nonperturbatively, we construct the mixed propagators constituting the cosmological correlators. A Mellin-Barnes approach helps isolate the pair of nonanalytic late-time branches carried by a heavy field. For a cubic isocurvature self-interaction, these branches contribute towards a squeezed bispectrum with a power-law envelope and logarithmic oscillations. The oscillation frequency encodes the heavy mass, while the amplitude and phase retain the full mixing dependence rather than a perturbative expansion. We perform the squeezed limit integral analytically and compare it with numerical results across representative masses and mixing strengths. At fixed mixing we derive the large-mass JWKB expansion of the squeezed correlator and its phase. We also give integral representations for the isocurvature Wightman function. The research idea was suggested by ARC, the calculation was performed by GPT, and the results were checked by the authors.

hep-th↗

arXiv2026

A Human-Centred Approach to Benchmarking LLMs for Parenting Advice

People are increasingly using large language models (LLMs) to seek advice, including for parenting. Parenting is a critical and socially sensitive domain. Thus, evaluating advice provided by LLMs requires indicators beyond aggregated information quality benchmarks to consider relational and behavioural elements of the responses. With a multi-dimensional rubric created by parenting experts, this paper evaluates 15 LLMs across 100 parenting scenarios in 2 languages (English and Chinese), using an LLM-as-a-judge method. Results show that aggregate scores can hide rubric item-specific weaknesses, models implicitly encourage different parenting styles, and language influences responses. We highlight the importance of evaluation output auditability and challenges involved in evaluating LLM-generated advice in domains like parenting. Our findings provide important insights for selecting LLMs for direct user engagement and the development of user-facing parenting advice applications.

cs.AI↗

arXiv2026

LongCat-Flash-Prover: Advancing Native Formal Reasoning via Agentic Tool-Integrated Reinforcement Learning

We introduce LongCat-Flash-Prover, a flagship 560-billion-parameter open-source Mixture-of- Experts (MoE) model that advances Native Formal Reasoning in Lean4 through agentic tool-integrated reasoning (TIR). We decompose the native formal reasoning task into three independent formal capabilities, i.e., auto-formalization, sketching, and proving. To facilitate these capabilities, we propose a Hybrid-Experts Iteration Framework to expand high-quality task trajectories, including generating a formal statement based on a given informal problem, producing a whole-proof directly from the statement, or a lemma-style sketch. During agentic RL, we present a Hierarchical Importance Sampling Policy Optimization (HisPO) algorithm, which aims to stabilize the MoE model training on such long-horizon tasks. It employs a gradient masking strategy that accounts for the policy staleness and the inherent train-inference engine discrepancies at both sequence and token levels. Additionally, we also incorporate theorem consistency and legality detection mechanisms to eliminate reward hacking issues. Extensive evaluations show that our LongCat-Flash-Prover sets a new state-of-the-art for open-weights models in both auto-formalization and theorem proving. Demonstrating remarkable sample efficiency, it achieves a 97.1% pass rate on MiniF2F-Test using only 72 inference budget per problem. On more challenging benchmarks, it solves 70.8% of ProverBench and 41.5% of PutnamBench with no more than 220 attempts per problem, significantly outperforming existing open-weights baselines.

cs.AI↗

arXiv2025

VitaBench: Benchmarking LLM Agents with Versatile Interactive Tasks in Real-world Applications

As LLM-based agents are increasingly deployed in real-life scenarios, existing benchmarks fail to capture their inherent complexity of handling extensive information, leveraging diverse resources, and managing dynamic user interactions. To address this gap, we introduce VitaBench, a challenging benchmark that evaluates agents on versatile interactive tasks grounded in real-world settings. Drawing from daily applications in food delivery, in-store consumption, and online travel services, VitaBench presents agents with the most complex life-serving simulation environment to date, comprising 66 tools. Through a framework that eliminates domain-specific policies, we enable flexible composition of these scenarios and tools, yielding 100 cross-scenario tasks (main results) and 300 single-scenario tasks. Each task is derived from multiple real user requests and requires agents to reason across temporal and spatial dimensions, utilize complex tool sets, proactively clarify ambiguous instructions, and track shifting user intent throughout multi-turn conversations. Moreover, we propose a rubric-based sliding window evaluator, enabling robust assessment of diverse solution pathways in complex environments and stochastic interactions. Our comprehensive evaluation reveals that even the most advanced models achieve only 30% success rate on cross-scenario tasks, and less than 50% success rate on others. Overall, we believe VitaBench will serve as a valuable resource for advancing the development of AI agents in practical real-world applications. The code, dataset, and leaderboard are available at https://vitabench.github.io/

cs.CL↗

arXiv2025

Meeseeks: A Feedback-Driven, Iterative Self-Correction Benchmark evaluating LLMs' Instruction Following Capability

The capability to precisely adhere to instructions is a cornerstone for Large Language Models (LLMs) to function as dependable agents in real-world scenarios. However, confronted with complex prompts, LLMs frequently encounter difficulties in fulfilling all specified requirements within a single response. Drawing inspiration from recent advancements in Chain-of-Thought (CoT) prompting and self-correction methodologies, we introduce Meeseeks (The name is inspired by Mr. Meeseeks from "Rick and Morty," a character renowned for efficiently accomplishing assigned tasks. See: https://en.wikipedia.org/wiki/Mr._Meeseeks), a fully automated iterative instruction-following benchmark equipped with an integrated feedback mechanism. Meeseeks identifies erroneous components in model responses and provides corresponding feedback accurately, thereby iteratively guiding the model toward self-correction. The dataset contains over 700 curated instances annotated by 32 distinct capability tags in Chinese and English. Extensive experimental results reveal that different state-of-the-art commercial and open-source LLMs exhibit vastly disparate performance, and even after 20 turns of iterative feedback-driven self-correction, nearly all models demonstrate suboptimal performance. We conducted comprehensive analysis from both macro and instance levels, uncovering numerous common issues prevalent in current state-of-the-art models, as well as several counterintuitive phenomena. We've open-sourced our work on https://github.com/ADoublLEN/Meeseeks.

cs.CL↗

arXiv2024

Geodesic completeness, cosmological bounces and inflation

The question of geodesic completeness of cosmological spacetimes has recently received renewed scrutiny. A particularly interesting result is the observation that the well-known Borde-Guth-Vilenkin (BGV) theorem may misdiagnose geodesically complete cosmologies. We propose a simple amendment to the BGV theorem which addresses such loopholes while retaining much of its generality. We give straightforward proofs of some recently offered conjectures concerning (generalized) Friedmann-Lema\^itre-Robertson-Walker spacetimes: geodesic completeness implies (i) the existence of a bounce, loitering phase or an emergent cosmology, and (ii) a phase of accelerated expansion with strictly increasing Hubble rate. Our results are purely kinematic and do not assume general relativity or energy conditions.

gr-qc↗