arXiv subjects

Yunfei Yu

Publications and source records attributed to Yunfei Yu.

9 recordsLinked to original sources

arXiv2026

Delegation Intelligence in Deep Search: A Controllable Framework for Disentangled Capability Diagnosis

Deep search is becoming a core capability of modern agent systems, yet it is typically evaluated solely based on end-to-end answer accuracy. This coupled evaluation paradigm entangles retrieval quality, long-context comprehension, evidence verification, and tool-use decisions, making it difficult to determine whether a model truly knows when and how to delegate information seeking to search. To this end: (1) We formalize this meta-capability as Delegation Intelligence in deep search and decompose it into complementary dimensions-Search Decision-Making (recognizing information insufficiency and deciding whether, when, and how to search) and Information Synthesis and Verification (aggregating evidence from multiple sources, judging source reliability, and synthesizing information under noisy, potentially adversarial conditions). (2) To enable disentangled and reproducible measurement, we develop a controllable synthesis pipeline built on document-grounded reverse engineering. This yields a general recipe for constructing controlled deep-search evaluations rather than a single fixed dataset. (3) As a concrete instantiation, we construct DelegSearchBench, together with a disentangled evaluation protocol that isolates each capability dimension by varying document composition and tool access. (4) Across representative models, we demonstrate that deep-search competence cannot be adequately characterized by final-answer accuracy alone...

cs.AI↗

arXiv2026

Faithful, Enriched, and Precise: Benchmarking Natural-Science Illustration Generation by T2I models

Scientific illustrations are essential tools for communicating research findings, especially in natural science, where they visualize complex concepts and processes. As Text-to-Image (T2I) models become increasingly capable, researchers have started to use them for scientific illustration generation. However, existing benchmarks often assess outputs at a holistic level, overlooking fine-grained elements, while scientific reasoning ability and output conciseness remain under-quantified. We introduce FEPBench, a benchmark built from carefully selected high-quality scientific illustrations across multiple disciplines and layout types. With the assistance of multimodal large language models (MLLMs) and human experts, we provide fine-grained atom set annotations and systematically evaluate T2I models along three dimensions: instruction faithfulness, reasoning enrichment, and semantic precision. Our evaluation further decomposes model performance across visual, textual, relation, and layout elements. Results show that even state-of-the-art (SOTA) closed-source models, such as GPT Image 2 and Nano Banana Pro, still suffer from text-rendering bottlenecks, limited reasoning enrichment, and difficulty balancing generation richness with precision. These findings provide practical guidance for improving and deploying T2I models in scientific illustration generation. Benchmark data, atom set annotations, and evaluation code will be released by us.

cs.CV↗

arXiv2026

GRIP: Geometric Refinement and Adaptive Information Potential for Data Efficiency

The performance of Large Language Models (LLMs) is increasingly governed by data efficiency rather than raw scaling volume. However, existing selection methods often decouple global distribution balancing from local instance selection, compromising the hierarchical integrity of the training set. We introduce \textbf{GRIP} (Geometric Refinement and Adaptive Information Potential), a framework that unifies these dimensions by modeling the corpus as an information-dense geometric space. GRIP employs a \textbf{Rapid Adaptation Probe (RAP)} to quantify the information potential of semantic clusters, dynamically re-allocating the sampling budget to regions with the highest representation deficits. Subsequently, we perform Intra-Cluster Selection using a \textbf{length-rectified geometric prior} to counteract embedding density artifacts and preserve long-tail logical sequences. Extensive evaluations on Mixture-of-Experts (MoE) models up to 300B tokens demonstrate that GRIP consistently outperforms state-of-the-art baselines, \textbf{surpassing the performance of models trained on $3\times$ larger uncurated datasets}. Our work establishes a robust geometric foundation for adaptive data curation in large-scale pre-training.

cs.CL↗

arXiv2026

UniGeM: Unifying Data Mixing and Selection via Geometric Exploration and Mining

The scaling of Large Language Models (LLMs) is increasingly limited by data quality. Most methods handle data mixing and sample selection separately, which can break the structure in code corpora. We introduce \textbf{UniGeM}, a framework that unifies mixing and selection by treating data curation as a \textit{manifold approximation} problem without training proxy models or relying on external reference datasets. UniGeM operates hierarchically: \textbf{Macro-Exploration} learns mixing weights with stability-based clustering; \textbf{Micro-Mining} filters high-quality instances by their geometric distribution to ensure logical consistency. Validated by training 8B and 16B MoE models on 100B tokens, UniGeM achieves \textbf{2.0$\times$ data efficiency} over a random baseline and further improves overall performance compared to SOTA methods in reasoning-heavy evaluations and multilingual generalization.

cs.LG↗

arXiv2023

Pressure-Modulated Structural and Magnetic Phase Transitions in Two-Dimensional FeTe: Tetragonal and Hexagonal Polymorphs

Two-dimensional (2D) Fe-chalcogenides with rich structures, magnetisms and superconductivities are highly desirable to reveal the torturous transition mechanism and explore their potential applications in spintronics and nanoelectronics. Hydrostatic pressure can effectively stimulate novel phase transitions between various ordered states and to plot the seductive phase diagram. Herein, the structural evolution and transport characteristics of 2D FeTe were systematically investigated under extreme conditions through comparing two distinct symmetries, i.e., tetragonal (t-) and hexagonal (h-) FeTe. We found that 2D t-FeTe presented the pressure-induced transition from antiferromagnetic to ferromagnetic states at ~ 3 GPa, corresponding to the tetragonal collapse of layered structure. Contrarily, ferromagnetic order of 2D h-FeTe was retained up to 15 GPa, evidently confirmed by electrical transport and Raman measurements. Furthermore, the detailed P-T phase diagrams of both 2D t-FeTe and h-FeTe were mapped out with the delicate critical conditions. We believe our results can provide a unique platform to elaborate the extraordinary physical properties of Fe-chalcogenides and further to develop their practical applications.

cond-mat.mtrl-sci↗

arXiv2023

Highly Anisotropic Elastic Properties of Suspended Black Arsenic Nanoribbons

Anisotropy, as an exotic degree of freedom, enables us to discover the emergent two-dimensional (2D) layered nanomaterials with low in-plane symmetry and to explore their outstanding properties and promising applications. 2D black arsenic (b-As) with puckered structure has garnered increasing attention these years owing to its extreme anisotropy with respect to the electrical, thermal, and optical properties. However, the investigation on mechanical properties of 2D b-As is still lacking, despite much effort on theoretical simulations. Herein, we report the highly anisotropic elastic properties of suspended b-As nanoribbons via atomic force microscope-based nanoindentation. It was found that the extracted Young's modulus of b-As nanoribbons exhibits remarkable anisotropy, which approximates to 72.2 +- 5.4 and 44.3 +- 1.4 GPa along zigzag and armchair directions, respectively. The anisotropic ratio reaches up to ~ 1.6. We expect that these results could lay a solid foundation for the potential applications of 2D anisotropic nanomaterials in the next-generation nanomechanics and optoelectronics.

cond-mat.mtrl-sci↗

arXiv2023

Phase-Modulated Elastic Properties of Two-Dimensional Magnetic FeTe: Hexagonal and Tetragonal Polymorphs

Two-dimensional (2D) layered magnets, such as iron chalcogenides, have emerged these years as a new family of unconventional superconductor and provided the key insights to understand the phonon-electron interaction and pairing mechanism. Their mechanical properties are of strategic importance for the potential applications in spintronics and optoelectronics. However, there is still lack of efficient approach to tune the elastic modulus despite the extensive studies. Herein, we report the modulated elastic modulus of 2D magnetic FeTe and its thickness-dependence via phase engineering. The grown 2D FeTe by chemical vapor deposition can present various polymorphs, i.e. tetragonal FeTe (t-FeTe, antiferromagnetic) and hexagonal FeTe (h-FeTe, ferromagnetic). The measured Young's modulus of t-FeTe by nanoindentation method showed an obvious thickness-dependence, from 290.9+-9.2 to 113.0+-8.7 GPa when the thicknesses increased from 13.2 to 42.5 nm, respectively. In comparison, the elastic modulus of h-FeTe remains unchanged. Our results could shed light on the efficient modulation of mechanical properties of 2D magnetic materials and pave the avenues for their practical applications in nanodevices.

cond-mat.mtrl-sci↗

arXiv2023

Experimental observation of highly anisotropic elastic properties of two-dimensional black arsenic

Anisotropic two-dimensional layered materials with low-symmetric lattices have attracted increasing attention due to their unique orientation-dependent mechanical properties. Black arsenic (b-As), with the puckered structure, exhibits extreme in-plane anisotropy in optical, electrical and thermal properties. However, experimental research on mechanical properties of b-As is very rare, although theoretical calculations predicted the exotic elastic properties of b-As, such as anisotropic Young's modulus and negative Poisson's ratio. Herein, experimental observations on highly anisotropic elastic properties of b-As were demonstrated using our developed in situ tensile straining setup based on the effective microelectromechanical system. The cyclic and repeatable load-displacement curves proved that Young's modulus along zigzag direction was ~1.6 times greater than that along armchair direction, while the anisotropic ratio of ultimate strain reached ~2.5, attributed to hinge structure in armchair direction. This study could provide significant insights to design novel anisotropic materials and explore their potential applications in nanomechanics and nanodevices.

cond-mat.mtrl-sci↗

arXiv2023

Intralayer Negative Poisson's Ratio in Two-Dimensional Black Arsenic by Strain Engineering

Negative Poisson's ratio as the anomalous characteristic generally exists in artificial architectures, such as re-entrant and honeycomb structures. The structures with negative Poisson's ratio have attracted intensive attention due to their unique auxetic effect and many promising applications in shear resistant and energy absorption fields. However, experimental observation of negative Poisson's ratio in natural materials barely happened, although various two-dimensional layered materials are predicted in theory. Herein, we report the anisotropic Raman response and the intrinsic intralayer negative Poisson's ratio of two-dimensional natural black arsenic (b-As) via strain engineering strategy. The results were evident by the detailed Raman spectrum of b-As under uniaxial strain together with density functional theory calculations. It is found that b-As was softer along the armchair than zigzag direction. The anisotropic mechanical features and van der Waals interactions play essential roles in strain-dependent Raman shifts and negative Poisson's ratio in the natural b-As along zigzag direction. This work may shed a light on the mechanical properties and potential applications of two-dimensional puckered materials.

cond-mat.mtrl-sci↗