Searcharxiv⌕ Search

arXiv subjects

Yifan Zhou

Publications and source records attributed to Yifan Zhou.

At least 37 records · Page 2Linked to original sources
arXiv2026

SemaPLC: A Project-Grounded, Verification-Gated Agent Harness for PLC Code Generation

Programmable logic controllers (PLCs) run industrial plants, and large language models can already generate independent program organization units (POUs) for them. Whether such logic integrates into an existing PLC project and then runs correctly has been checked only in limited tests. We present \textsc{SemaPLC}, a project-grounded and verification-gated agent harness assembled from conventional tools but governed by a strict completion rule. Rather than stopping when the model judges its own output adequate, \textsc{SemaPLC} declares a task complete only when logged external checks confirm it. Those checks cover the specification, the compilation, and the behavior on a live runtime. On 117 independent-POU tasks matching existing benchmarks, it attains the highest strict verified pass rate on all seven models (72.6\% mean). On a project-context track of 65 tasks whose generated logic must compile and run inside a real project, it attains the highest mean on integrated compilation, static behavior, and dynamic behavior. Of the three layers, dynamic behavior is the most revealing. We measure it by deploying the generated and the reference logic to a live PLC runtime and comparing their executed traces. All methods fall within 10 static points of one another, whereas dynamic scores separate them sharply, from 22.4 to 31.4 for the baselines against 52.2 for \textsc{SemaPLC}. Overall, our verification-gated harness raises the mean at every layer and most sharply at runtime. Execution, not static scoring, is the faithful test of whether generated control logic actually works. \textsc{SemaPLC} is open-sourced at https://github.com/midea-ai/SemaPLC.

cs.SE↗
arXiv2026

StagedWorkspace: A Versioned Workspace for Knowledge-Work Agents

AI agents increasingly perform knowledge work (i.e., produce and modify persistent digital artifacts such as code repositories, documents, spreadsheets, slides, reports), yet the parsed views they search, the native files they edit, the changes they review, and the artifacts they submit can refer to different versions of the same work product. We formulate this as a workspace-state contract: every view should be explicitly tied to a version of the evolving workspace state. Coding agents partly address this need through repository contracts for search, diffs, and tests, whereas an analogous contract is less explicit for PDFs, spreadsheets, slides, notebooks, and mixed-format project folders. We propose StagedWorkspace, a versioned workspace for knowledge-work agents. The workspace binds parsed records and review diffs to content hashes of the native files as they change. In fixed-harness ablations on OfficeQA Pro and APEX-Agents, dual parsed/native access has the highest point estimate for every tested model; relative to the more limiting single view, it improves OfficeQA Pass@1 by 8.3-12.1 points and APEX mean rubric score by 4.7-9.2 points. SW-AGENT scores 63.9% with Gemini 3.1 Pro on OfficeQA and 42.1 with GPT-5.4 Nano on APEX, compared with published same-model scores of 29.3% and 25.5, respectively. A paired review-axis ablation on 57 file-editing tasks further finds higher observed scores when diffs are visible. These results identify workspace state as an experimental variable in knowledge-work agents and motivate benchmarks that score evidence, staged edits, and submitted artifacts as explicit state transitions.

cs.AI↗
arXiv2026

Thinking with Gaze: Sequential Eye-Tracking as Visual Reasoning Supervision for Medical VLMs

Vision--language models (VLMs) process images as visual tokens, yet their intermediate reasoning is often carried out in text, which can be suboptimal for visually grounded radiology tasks. Radiologists instead diagnose via sequential visual search; eye-tracking captures this process as time-ordered gaze trajectories that reveal how evidence is acquired over time. We use eye-gaze as supervision to guide VLM reasoning by introducing a small set of dedicated gaze tokens. These tokens are trained to predict gaze-selected image patch indices in temporal order, encouraging the model to follow human-like evidence acquisition and integration. Experiments on MIMIC-EYE and multiple external zero-shot benchmarks show consistent gains over baselines, achieving state-of-the-art in-domain performance and improved out-of-domain robustness. These results highlight temporally ordered gaze as an effective supervision signal for learning visually grounded medical reasoning.

cs.CV↗
arXiv2026

Revolutionizing Finance with LLMs: An Overview of Applications and Insights

In recent years, Large Language Models (LLMs) like ChatGPT have seen considerable advancements and have been applied in diverse fields. Built on the Transformer architecture, these models are trained on extensive datasets, enabling them to understand and generate human language effectively. In the financial domain, the deployment of LLMs is gaining momentum. These models are being utilized for automating financial report generation, forecasting market trends, analyzing investor sentiment, and offering personalized financial advice. Leveraging their natural language processing capabilities, LLMs can distill key insights from vast financial data, aiding institutions in making informed investment choices and enhancing both operational efficiency and customer satisfaction. In this study, we provide a comprehensive overview of the emerging integration of LLMs into various financial tasks. Additionally, we conducted holistic tests on multiple financial tasks through the combination of natural language instructions. Our findings show that GPT-4 effectively follow prompt instructions across various financial tasks. This survey and evaluation of LLMs in the financial domain aim to deepen the understanding of LLMs' current role in finance for both financial practitioners and LLM researchers, identify new research and application prospects, and highlight how these technologies can be leveraged to solve practical challenges in the finance industry.

cs.CL↗
arXiv2026

Same Targets, Different Computation: How Post-Training Divides Work Across Model Layers

A late-layer change learned during post-training may work on the base model's earlier state, or it may depend on earlier computation learned with it. We distinguish these cases with a four-cell diagnostic that crosses base or descendant upstream states with base or descendant late stacks. A large late-stack effect need not imply strong upstream dependence. On math prompts, OpenMath2's late stack changes the target margin by +3.43 logits after base upstream state and +3.28 after its own, giving a near-zero interaction. Several instruction-following descendants of the same Llama-3.1-8B base show greater dependence, while controlled code and biomedical continuation-training runs sit near zero on the common support; seven released descendants span -0.54 to +2.20 logits. Because those checkpoints differ in many ways, we then isolate one training property. Two LoRA fine-tunes learn identical target responses requested either by familiar natural-language instructions or by newly learned nonce codes. Changing only this cue-to-response relation increases upstream dependence by +5.56 logits on Qwen3-4B and +4.18 on Llama-3.1-8B, with a positive paired shift in all six model-by-seed runs. The interaction is also positive in all five released base/instruction pairs we test, and late-stack replacement changes the full-vocabulary argmax in about half of events. Post-training can therefore organize the same target behavior with different dependencies between earlier and later computation. The diagnostic measures local next-token compatibility, not free-running component transplantation.

cs.LG↗
arXiv2026

One-Step Epitaxial Access to Rhombohedral Graphene Flat-Band States on Step-Bunched SiC

Rhombohedral graphene multilayers provide a moiré-free platform for correlated and topological flat-band physics, but direct, transfer-free epitaxial access to thickness-tunable multilayers remains limited. Here we report a one-step graphitization route on 4$^\circ$ off-axis 4H-SiC, in which high-temperature flash annealing simultaneously drives self-organized step bunching and multilayer graphene formation. Atomic-resolution cross-sectional scanning transmission electron microscopy identify local ABC registry and distinguish rhombohedral from Bernal stacking. The thickness is tuned from bilayer to more than twenty layers by varying single parameter, the annealing temperature. Angle-resolved photoemission spectroscopy directly tracks the thickness-dependent evolution from interface-dominated low-energy states toward pronounced near-Fermi-level flat-band spectral weight in thick multilayers. Low-temperature scanning tunneling microscopy and spectroscopy on a 17-layer film further reveal a 13.4 meV low-energy spectral reconstruction and a $\sqrt{3} \times \sqrt{3}$ Kekulé-like modulation, providing microscopic signatures consistent with an intervalley-mixed electronic texture. This one-step, transfer-free approach establishes step-bunched SiC as an epitaxial platform that links stacking engineering with moiré-free correlated flat-band electronic states.

cond-mat.mtrl-sci↗
arXiv2026

The JWST weather report: Unravelling the atmospheric variability of isolated worlds using principal component analysis

Brown dwarf variability directly probes atmospheric dynamics beyond the Solar System, and recent JWST time-resolved spectroscopy has opened a new window into these processes. Principal component analysis (PCA) offers a data-driven framework to identify the dominant, independent patterns of spectral variability of variable targets without relying on prior atmospheric assumptions. SIMP 0136 is a young, T2.5, brown dwarf at the planetary-mass boundary, making it an ideal analogue for directly imaged exoplanets. We analysed one rotation of JWST/NIRSpec PRISM time-series spectroscopy to investigate the drivers of its variability using PCA. Two principal components are sufficient to reduce the residual spectra to the propagated noise floor, indicating that they capture the detectable coherent spectroscopic variability. The leading principal component captures broadband variability consistent with temperature changes, while the second traces chromatic variability linked to vertical cloud structure. The dominance of two components implies that the spectra can be described as mixtures of three distinct atmospheric states, whose relative contributions we mapped as a function of rotational phase. The observed spectra are described as evolving linear combinations of these states, indicating that the variability arises from the changing visibility of spatially distinct atmospheric regions. By projecting Sonora Diamondback forward models into the same principal component space, we found that the principal components capture a large fraction of the model variance, demonstrating that the same physical processes that govern SIMP-0136's observed variability also capture much of the model grid's variation. Our results establish PCA as a computationally efficient, physically interpretable framework for analysing JWST time-resolved spectroscopy of substellar atmospheres.

astro-ph.EP↗
arXiv2026

Photometric Variability and Rotation of Beta Pictoris b from JWST NIRCam Coronagraphic Imaging

We report the detection of photometric variability in the directly imaged super-Jupiter $β$ Pictoris b. Using JWST NIRCam dual-band coronagraphic imaging, we conducted a 16-hour continuous photometric monitoring campaign in the F210M and F410M filters. We developed and validated a time-series photometry framework that combines PSF subtraction, principal component analysis for systematic noise removal, and injection-and-recovery tests to confirm signal fidelity. Both light curves show consistent sinusoidal variability at $\sim$5$σ$ and $\gg 5σ$ significance in the F210M and F410M bands, respectively. A joint sinusoidal fit yields a rotation period of $P_{\rm rot} = 9.00 \pm 0.13$ hr and variability amplitudes of $0.85 \pm 0.07\%$ and $0.89 \pm 0.04\%$ in F210M and F410M, respectively. The near-identical amplitudes and periods in both bands confirm a common astrophysical origin in a heterogeneous atmosphere. Combining $P_{\rm rot}$ with the previously measured projected rotational velocity, we constrain the line-of-sight spin axis inclination of $β$ Pic b. The result favors an equator-on viewing geometry, consistent with line-of-sight spin-orbit alignment: the planetary spin axis, orbital plane, debris disk, and stellar equator are all mutually aligned. This stands in sharp contrast to the large obliquities of wide-separation companions that are likely formed via gravitational fragmentation. Together with the system's young age, this observation provides independent dynamical evidence that $β$ Pic b formed via core accretion. This result constitutes the first detection of rotational modulation in a close-in, high-contrast exoplanet that likely formed via core accretion, demonstrating that time-series coronagraphic imaging with JWST opens a powerful new window onto the rotation, atmospheric dynamics, and spin-orbit architecture of this population.

astro-ph.EP↗
arXiv2026

Trainable Log-linear Sparse Attention for Efficient Diffusion Transformers

Diffusion Transformers (DiTs) set the state of the art in visual generation, yet their quadratic self-attention cost fundamentally limits scaling to long token sequences. Recent Top-K sparse attention approaches reduce the computation of DiTs by compressing tokens into block-wise representation and selecting a small set of relevant key blocks, but still suffer from (i) quadratic selection cost on compressed tokens and (ii) increasing K required to maintain model quality as sequences grow. We identify that their inefficiency is due to the single-level design, as a single coarse level is insufficient to represent the global structure. In this paper, we introduce Log-linear Sparse Attention (LLSA), a trainable sparse attention mechanism for extremely long token sequences that reduces both selection and attention costs from quadratic to log-linear complexity by utilizing a hierarchical structure. LLSA performs hierarchical Top-K selection, progressively adopting sparse Top-K selection with the indices found at the previous level, and introduces a Hierarchical KV Enrichment mechanism that preserves global context while using fewer tokens of different granularity during attention computation. To support efficient training, we develop a high-performance GPU implementation that uses only sparse indices for both the forward and backward passes, eliminating the need for dense attention masks. We evaluate LLSA on high-resolution pixel-space image generation without using patchification and VAE encoding. LLSA accelerates attention inference by 28.27x and DiT training by 6.09x on 256x256 pixel token sequences, while maintaining generation quality. The results demonstrate that LLSA offers a promising direction for training long-sequence DiTs efficiently. Code is available at: https://github.com/SingleZombie/LLSA

cs.CV↗
arXiv2026

Direct Imaging Discovery of Giant Exoplanet $β$ Pictoris d: A Decade-Long Game of Hide-and-Seek

We report the direct imaging discovery of a third exoplanet in the $β$ Pictoris system. We detect $β$ Pictoris d ($β$ Pic d) in non-coronagraphic observations obtained with VLT/ERIS as well as multi-epoch archival datasets from JWST/NIRCam and VLT/SPHERE. Astrometric measurements over an 11-year baseline demonstrate that it is consistent with a gravitationally-bound source with orbital motion. Joint multi-planet orbit fits of all three planets in the system yield a semi-major axis of $26.0^{+2.2}_{-6.1}$ au and inclination $89.0^{+0.7}_{-0.6}$ deg for planet d. $β$ Pic d has a larger orbital semi-major axis than the other known planets in the system, but is coplanar with the inner two planets, and its orbit is consistent with sculpting the inner edge of the debris disk. $β$ Pic d has a contrast of $ΔL^{\prime}=12.11\pm0.15$ mag, with colors and luminosity that closely match those of 51 Eri b, another exoplanet in the $β$ Pictoris moving group. Its VLT/ERIS and JWST/NIRCam colors are distinct from those of free-floating planetary-mass objects of a similar age and temperature. Its red $F410M-F444W$ color indicates strong CO$_2$ absorption in its atmosphere and suggests significant enhancement in metals compared to free-floating objects. From the ATMO hot-start evolutionary models, we estimate an effective temperature of $600^{+45}_{-60}$ K and mass of $2.4\pm0.6$ $M_{\rm Jup}$, which also closely matches similar estimates for 51 Eri b. $β$ Pic d is among the lowest-mass exoplanets imaged from the ground. This discovery highlights the deep sensitivity achievable with ground-based imaging in the mid-infrared and the discovery potential of future high-contrast observations with the Extremely Large Telescope.

astro-ph.EP↗
arXiv2026

Ideas Have Genomes: Benchmarking Scientific Lineage Reasoning and Lineage-Grounded Idea Generation

Scientific ideas rarely start from a blank page. They inherit mechanisms, repair known limitations, and recombine pieces of earlier work, much like biological genomes. Current benchmarks still say little about whether AI systems can follow this inheritance structure. We present IdeaGene-Bench (IG-Bench), a benchmark for scientific lineage reasoning and lineage-grounded idea generation. IG-Bench is organized around the IdeaGene framework: each paper or proposal is represented as a set of minimal, typed, evidence-grounded Idea Genome objects, and a GenomeDiff aligns these objects to record inheritance, mutation, loss, external import, and novel insertion under six operational evolutionary dynamics. The benchmark contains 1,961 golden lineage traces, 1,085 curated Idea Genome objects, and 920 pairwise GenomeDiff records across 10 scientific domains. It supports two evaluations. IG-Exam (42 task types, 1,029 instances) tests closed-form lineage reasoning across Idea Genome abstraction, inheritance tracing, evolutionary reasoning, and lineage verification. IG-Arena evaluates generation with a lineage-conditioned Population-Evolution Score(PES), asking whether a proposal can be inserted as a coherent descendant of a given lineage population: it should inherit the right Idea Genome objects, vary meaningfully from nearby work, and offer selection value for future research. Experiments on 14 LLM-based scientists expose a compositional bottleneck. The strongest system reaches only 27.3% exact accuracy on lineage reasoning, and structured lineage context reshuffles system rankings rather than helping every participant uniformly.

cs.AI↗
arXiv2026

ResearchClawBench: A Benchmark for End-to-End Autonomous Scientific Research

AI coding agents are increasingly used for scientific work, but their end-to-end autonomous research capability remains difficult to verify. We present ResearchClawBench, a benchmark for evaluating autonomous scientific research across 40 tasks from 10 scientific domains. Each task is grounded in a real published paper, provides related literature and raw data, and hides the target paper during evaluation. Expert-curated multimodal rubrics decompose the target scientific artifacts into weighted criteria, enabling evaluation of target-paper-level re-discovery while leaving room for new discovery. We evaluate seven autonomous research (auto-research) agents under a unified protocol and seventeen native LLMs through the lightweight ResearchHarness. Current systems remain far from reliable re-discovery: the strongest autonomous agent, Claude Code, averages 21.5, and the strongest ResearchHarness LLM, Claude-Opus-4.7, averages 20.7, with an LLM frontier mean of only 26.5. Error analysis shows that failures concentrate in experimental protocol mismatch, evidence mismatch, and missing scientific core. ResearchClawBench provides a reproducible evaluation frontier for measuring progress toward autonomous scientific research.

cs.LG↗
arXiv2026

SciFig: Towards Automating Editable Figure Generation for Scientific Papers

High-quality methodology figures are central to scientific communication, yet they remain difficult and time-consuming to create. Such figures must distill a method's components and information flow into a clear, revisable diagram as the paper evolves. Existing methodology diagram automation systems typically face a trade-off between editability and visual quality: TikZ- or SVG-based methods produce editable structured outputs but often lack the richness of human-designed figures, while image-generation models produce polished raster outputs that are difficult to revise. We introduce SciFig, an end-to-end multi-agent framework for generating visually rich and fully editable methodology figures from scientific text. SciFig decomposes figure generation into planning, layout synthesis, component rendering, and iterative refinement, producing XML figures that can be edited in standard diagramming tools and refined through human or VLM feedback. We also introduce SciFig-Bench, a human-verified benchmark of 435 author-drawn methodology figures from 37 arXiv domains and 15 top-tier AI/ML venues, and SciFig-Eval, a four-axis evaluation protocol for measuring figure quality. Across seven single-agent and agentic baselines, SciFig achieves the best performance on all four SciFig-Eval axes and generates editable figures in about 10 minutes on average. Qualitative examples further show that SciFig can generalize beyond methodology figures to teaser diagrams and statistical plots. Dataset and code are available at: https://shramanpramanick.github.io/SciFig/.

cs.AI↗
arXiv2026

Nautilus Space Observatory: The Evolution of Planets and their Atmospheres

We are just beginning to explore the billion-year evolution from nascent planets in disks to mature planetary systems. Recent discoveries hint at demographic and atmospheric differences between young planets and their Gyr-old counterparts, but current facilities are limited - particularly in their ability to conduct statistical atmospheric studies over a broad period range. This white paper outlines compelling science achievable with the Nautilus Space Observatory, a proposed constellation of large-diameter space telescopes. We identify four primary scientific objectives: (1) determining the timescales over which planets evolve into sub-Neptunes and super-Earths; (2) tracking the temporal evolution of atmospheric mass-loss rates; (3) characterizing the evolution of the atmospheric mean molecular weight and C/O ratio; and (4) identifying the emergence of Helium-dominated worlds. Answering these questions requires the high spatial resolution, broad-wavelength coverage, large effective area, and parallelized multiple units that Nautilus provides. By isolating the physical processes that govern the evolution of planets and their atmospheres, these science objectives directly support NASA's Cosmic Origins and Exoplanet Exploration Programs.

astro-ph.IM↗
arXiv2026

Toward Vibe Medicine: A Self-Evolving Multi-Agent Framework for Clinical Decision Support

In recent years, the advances of large language models and autonomous agents have revolutionized the healthcare field, facilitating diagnosis and improving treatment results. However, most existing AI systems rely on pre-trained knowledge and predefined pipelines, which struggle to learn dynamically from the interactive chat session history that contains patient outcomes and past failures. To address this limitation, we propose VIBEMed, a multi-agent framework with a built-in self-evolution mechanism and architecture-level safety sandbox for robust clinical decision support. The system integrates three specialized agents, including a Clinical Diagnostic Agent (CDA) for hypothesis generation, a Therapeutic Execution Agent (TEA) for treatment planning, and a Clinical Evolution Manager Agent (CEMA) that distills longitudinal clinical feedback into reusable knowledge, transforming multimodal patient information into personalized medical decisions. Through self-evolution mechanism, the framework enables iterative updates across memory, model behavior, and decision strategies, allowing the system to improve over time. Experimental results show that VIBEMed demonstrates superior performance through its evolving mechanism in complex clinical cases, particularly in tasks that require integrated decision-making and longitudinal planning. The framework also supports reliable end-to-end decisions in challenging scenarios such as oncology treatment planning, highlighting its feasibility in real-world clinical contexts. Overall, VIBEMed provides a practical path beyond static AI systems toward adaptive, experience-driven clinical decision support, demonstrating the value of combining multi-agent collaboration with continuous evolution for advancing precision medicine.

cs.AI↗
arXiv2026

Ultraviolet Imaging of SR 12 c with HST/WFC3: Accretion and Variability of a Giant Planet at the End Stages of Growth

Many details of the gas accretion phase during giant planet formation remain untested. We present new 0.2$\unicode{x2013}$0.7 $μ$m UV-through-red optical imaging of the young, wide-orbit planetary-mass companion SR 12 c from the Wide Field Camera 3 (WFC3) instrument on board the Hubble Space Telescope. SR 12 c exhibits strong accretion-related continuum excess blueward of $\sim$5000 $\unicode{x212B}$ and clear signs of the Balmer jump at 3646 $\unicode{x212B}$. We derive a total accretion luminosity of 1.65 $\pm$ $0.19 \times 10^{-5} L_{\odot}$ and a mass accretion rate of 8 $\pm$ $2\times 10^{-12}$ M$_{\odot}$ yr$^{-1}$. Based on its mass and age, SR 12 c will not grow by an appreciable amount at its current accretion rate; it is at the end stages of assembly. No accretion variability is evident between the two epochs of the WFC3 observations spanning a month-long baseline, but the H$α$ emission line strength decreases by 90% compared to the reported flux from five years earlier. Combined with previous observations of SR 12 c, we assemble one of the most complete spectral energy distributions of a young giant planet to date, spanning the UV through sub-mm wavelengths (0.2$\unicode{x2013}$880 $μ$m). This adds SR 12 c to the small yet growing sample of planets with detailed accretion and disk constraints, which together are beginning to establish the diversity of timescales and physical processes governing the formation of giant planets.

astro-ph.EP↗
arXiv2026

HyperTool: Beyond Step-Wise Tool Calls for Tool-Augmented Agents

Tool-augmented LLM agents commonly rely on step-wise atomic tool calls, where each invocation, observation, and value transfer is exposed in the main reasoning trace. This creates an \emph{execution-granularity mismatch}: locally deterministic tool workflows are unfolded into repeated model-visible decisions, consuming context and forcing the model to manage low-level dataflow in the trace. We introduce \textbf{HyperTool}, a unified executable MCP-style tool interface that changes the model-visible unit of tool execution. A model invokes HyperTool with a code block that can call existing tools through their original schemas, manipulate returned values, and pass intermediate results locally, folding deterministic tool subroutines into a single outer call. To train models to use this interface, we synthesize HyperTool-format trajectories from cross-tool compositional tasks and verify them in real MCP environments. On MCP-Universe, HyperTool improves average accuracy from 15.69\% to 35.29\% on Qwen3-32B and from 9.93\% to 33.33\% on Qwen3-8B, and surpass GPT-OSS and Kimi-k2.5 on average accuracy, showing that our HyperTool can substantially improve multi-step tool use.

cs.CL↗
arXiv2026

MLEvolve: A Self-Evolving Framework for Automated Machine Learning Algorithm Discovery

Large language model (LLM) agents are increasingly applied to long-horizon tasks such as scientific discovery and machine learning engineering (MLE), where sustained self-evolution becomes a key capability. However, existing MLE agents suffer from inter-branch information isolation, memoryless search, and lack of hierarchical control, which together hinder long-horizon optimization. We present MLEvolve, an LLM-based self-evolving multi-agent framework for end-to-end machine learning algorithm discovery. By extending tree search to Progressive MCGS, MLEvolve enables cross-branch information flow through graph-based reference edges and gradually shifts the search from broad exploration to focused exploitation with an entropy-inspired progressive schedule. To allow the agent to evolve with accumulated experience, we introduce Retrospective Memory, which combines a cold-start domain knowledge base with a dynamic global memory for task-specific experience retrieval and reuse. For stable long-horizon iteration, we further decouple strategic planning from code generation with adaptive coding modes. Evaluation on MLE-Bench shows that MLEvolve achieves state-of-the-art performance across multiple dimensions including average medal rate and valid submission rate under a 12-hour budget (half the standard runtime). Moreover, MLEvolve also outperforms specialized algorithm discovery methods including AlphaEvolve on mathematical algorithm optimization tasks, demonstrating strong cross-domain generalization. Our code is available at https://github.com/InternScience/MLEvolve.

cs.AI↗