papers checked against Recognition

60 of 33950 papers

• Ghost in the Minecraft: Generally Capable Agents for Open-World Environments via Large Language Models with Text-based Knowledge and Memory

  cs.AI · 2023-05-25 · 3 theorem links

  GITM uses LLMs to generate action plans from text knowledge and memory, enabling agents to complete long-horizon Minecraft tasks at much higher success rates than prior RL methods.

• Language Is Not All You Need: Aligning Perception with Language Models

  cs.CL · 2023-02-27 · 3 theorem links

  Kosmos-1 shows strong zero-shot and few-shot results on language tasks, image captioning, visual QA, OCR-free document understanding, and image recognition guided by text instructions.

• Mass-type invariants in the presence of a cosmological constant

  math.DG · 2026-03-02 · 3 theorem links

  New mass invariants for positive cosmological constant initial data characterize de Sitter spacetime and yield positive mass and Penrose inequalities.

• ARC-AGI-2: A New Challenge for Frontier AI Reasoning Systems

  cs.AI · 2025-05-17 · 3 theorem links

  ARC-AGI-2 adds a larger, more complex set of tasks to the original ARC-AGI benchmark to give finer-grained measurement of fluid intelligence in AI.

• Q-Align: Teaching LMMs for Visual Scoring via Discrete Text-Defined Levels

  cs.CV · 2023-12-28 · 3 theorem links

  Q-Align trains LMMs on discrete text-defined levels for visual scoring, achieving SOTA on IQA, IAA, and VQA while unifying the tasks in OneAlign.

• What Does Flow Matching Bring To TD Learning?

  cs.LG · 2026-03-04 · 3 theorem links

  Flow matching critics outperform monolithic ones in RL by 2x performance and 5x sample efficiency via test-time error recovery through integration and multi-point velocity supervision that preserves feature plasticity.

• Seed Diffusion: A Large-Scale Diffusion Language Model with High-Speed Inference

  cs.CL · 2025-08-04 · 3 theorem links

  Seed Diffusion Preview is a discrete diffusion language model that reaches 2146 tokens per second inference on H20 GPUs with competitive code benchmark performance, establishing a new speed-quality Pareto frontier.

• Regge trajectories from the adjoint sector of Matrix Quantum Mechanics

  hep-th · 2026-03-04 · 3 theorem links

  At criticality in the adjoint sector of matrix quantum mechanics, the spectrum exhibits Regge trajectories Δ² ~ n/α' interpreted as oscillatory excitations of short folded open strings in the dual 2D string theory.

• SpinQuant: LLM quantization with learned rotations

  cs.LG · 2024-05-26 · 3 theorem links

  SpinQuant learns optimal rotations to enable accurate 4-bit quantization of LLM weights, activations, and KV cache, reducing the zero-shot gap to full precision to 2.9 points on LLaMA-2 7B.

• AutoUE: Automated Generation of 3D Games in Unreal Engine via Multi-Agent Systems

  cs.HC · 2026-03-07 · 3 theorem links

  AutoUE is a multi-agent system that uses retrieval-augmented generation and automated play-testing to generate end-to-end 3D games in Unreal Engine.

• VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning

  cs.CV · 2021-05-11 · 3 theorem links

  VICReg prevents collapse in self-supervised image embeddings via explicit variance, invariance, and covariance regularization and matches state-of-the-art downstream performance.

• NePPO: Near-Potential Policy Optimization for General-Sum Multi-Agent Reinforcement Learning

  cs.LG · 2026-03-07 · 3 theorem links

  NePPO learns a player-independent potential function via a novel objective whose minimization yields an approximate Nash equilibrium for general-sum multi-agent games.

• Senna: Bridging Large Vision-Language Models and End-to-End Autonomous Driving

  cs.CV · 2024-10-29 · 3 theorem links

  Senna decouples language-based high-level planning from an LVLM with low-level trajectory prediction from an E2E model, reporting 27% lower planning error and 33% lower collisions after pre-training on DriveX and fine-tuning on nuScenes.

• Circle packing and Riemann uniformization of random planar maps in an ergodic scale-free environment

  math.PR · 2026-03-06 · 3 theorem links

  Infinite planar maps in ergodic scale-free environments converge at large scales to their circle-packing and Riemann-uniformization embeddings under suitable moment and connectivity conditions.

• Exact Is Easier: Credit Assignment for Cooperative LLM Agents

  cs.LG · 2026-03-06 · 3 theorem links

  C3 achieves exact unbiased credit assignment in cooperative LLM agents by exploiting deterministic observable histories for counterfactual sampling, outperforming approximate baselines across six benchmarks.

• Methodological opportunities for mitigating climate change in complex food systems

  physics.soc-ph · 2026-03-10 · 3 theorem links

  A shared methodological toolkit from multiple scientific domains is outlined to measure structure versus randomness in food systems, predict transitions, and inform climate-adaptive redesigns.

• Neural Ordinary Differential Equations

  cs.LG · 2018-06-19 · 3 theorem links

  Neural networks are redefined as continuous dynamical systems by learning the derivative of the hidden state with a neural network and integrating it with an ODE solver.

• AR-VLA: True Autoregressive Action Expert for Vision-Language-Action Models

  cs.RO · 2026-03-10 · 3 theorem links

  AR-VLA introduces a standalone autoregressive action expert with long-lived memory that generates context-aware continuous actions for VLAs, replacing chunk-based heads with smoother trajectories and maintained task success.

• Seed-TTS: A Family of High-Quality Versatile Speech Generation Models

  eess.AS · 2024-06-04 · 3 theorem links

  Seed-TTS models produce speech matching human naturalness and speaker similarity, with added controllability via self-distillation and reinforcement learning.

• Hankel Determinants from Quadratic Orthogonal Pairs for Hyperelliptic Functions and Their Applications

  nlin.SI · 2026-03-12 · 3 theorem links

  Quadratic orthogonal pairs resolve the mismatch problem for Hankel determinants on hyperelliptic curves and supply closed-form solutions for bilateral Somos-4 and Somos-5 recurrences.

• Measuring Primitive Accumulation: An Information-Theoretic Approach to Capitalist Enclosure in PIK2, Indonesia

  physics.soc-ph · 2026-03-14 · 3 theorem links

  Satellite data projected onto a Marxian simplex shows a 0.405 rad/yr transformation pulse, 38-46 year absorption times into built land, and percolation below random thresholds indicating planned rather than stochastic urban growth in PIK2.

• Fixed point theorem for cluster modular groups

  math.GT · 2026-03-15 · 3 theorem links

  Finite subgroups of cluster modular groups have fixed points in the positive cluster varieties under a condition verified for all finite mutation types except X_7.

• CodeXGLUE: A Machine Learning Benchmark Dataset for Code Understanding and Generation

  cs.SE · 2021-02-09 · 3 theorem links

  CodeXGLUE supplies a standardized collection of 10 code-related tasks, 14 datasets, an evaluation platform, and BERT-, GPT-, and encoder-decoder-style baselines.

• Linearized Attention Cannot Enter the Kernel Regime at Any Practical Width

  cs.LG · 2026-03-13 · 3 theorem links

  Linearized attention requires impractically large widths m = Ω(κ_d(G)^6 n log n) to enter the NTK regime, exceeding 10^24 for MNIST and 10^29 for CIFAR-10.

• Lingshu: A Generalist Foundation Model for Unified Multimodal Medical Understanding and Reasoning

  cs.CL · 2025-06-08 · 3 theorem links

  Lingshu is a medical-specialized multimodal LLM that outperforms prior open-source models on multimodal QA, text QA, and report generation after training on a large curated dataset of medical knowledge.

• A new model for the quantum mechanics of the Hydrogen atom

  math-ph · 2026-03-16 · 3 theorem links

  The model on the Lorentzian cone with algebraic operators reproduces the standard hydrogen atom energy spectrum and wavefunction solutions in a Schwartz subspace.

• The Dimension of Integral Self-Affine Sets via Fractal Perturbations: The Box and the Hausdorff Dimensions, Ergodic Measures

  math.DS · 2026-03-15 · 3 theorem links

  A fractal perturbation technique yields the dimensions of integral self-affine sets and proves existence of full-dimensional ergodic measures on their torus versions.

• SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution

  cs.SE · 2025-02-25 · 3 theorem links

  SWE-RL uses RL on software evolution data to train LLMs achieving 41% on SWE-bench Verified with generalization to other reasoning tasks.

• Projection and Invariance in Scientific Explanation

  physics.hist-ph · 2026-03-17 · 3 theorem links

  Projection is the central mechanism of scientific representation: a mapping that partitions states into equivalence classes, omits internal variation, and thereby reveals invariants that would otherwise remain inaccessible.

• REALM: Retrieval-Augmented Language Model Pre-Training

  cs.CL · 2020-02-10 · 3 theorem links

  REALM augments language-model pre-training with an unsupervised retriever over Wikipedia documents and reports 4-16% absolute gains on open-domain QA benchmarks over prior implicit and explicit knowledge methods.

• Inference-Time Intervention: Eliciting Truthful Answers from a Language Model

  cs.LG · 2023-06-06 · 3 theorem links

  ITI shifts activations in limited attention heads using directions found from a few hundred examples, raising Alpaca's TruthfulQA truthfulness from 32.5% to 65.1% while allowing tunable tradeoff with helpfulness.

• On the Duality of Coverings in Hilbert Geometry

  math.MG · 2026-03-19 · 3 theorem links

  Hilbert-geometry covering numbers satisfy polarity duality: N^H_K(G, α) is bounded above and below by c^{±d} times N^H_{G°}(K°, α) for an absolute constant c.

• Growing Alphabets in Canonical Shuffle Experiments: Likelihood-Ratio Laws, Estimation Bounds, and Low-Budget Equivariant Design

  cs.IT · 2026-03-18 · 3 theorem links

  Alphabet growth improves canonical shuffled privacy precisely when the worst pairwise likelihood-ratio law collapses to a point mass, with sharp chi-squared bounds and optimal low-budget equivariant designs.

• Quantum Structures as Generative Scores: Partition Logic, Generative Logic, and Aesthetic Form

  quant-ph · 2026-03-19 · 3 theorem links

  Partition logic is converted into Prolog generative grammars to produce modular artifacts like the Quantum Square, keeping logical structure independent of its realization for design and communication uses.

• ReCogDrive: A Reinforced Cognitive Framework for End-to-End Autonomous Driving

  cs.CV · 2025-06-09 · 3 theorem links

  ReCogDrive unifies VLM scene understanding with a diffusion planner reinforced by DiffGRPO to reach state-of-the-art results on NAVSIM and Bench2Drive benchmarks.

• MRKL Systems: A modular, neuro-symbolic architecture that combines large language models, external knowledge sources and discrete reasoning

  cs.CL · 2022-05-01 · 3 theorem links

  MRKL is a modular neuro-symbolic architecture that integrates LLMs with external knowledge and discrete reasoning to overcome limitations of pure neural language models.

• Interpretable EEG Microstate Discovery via Variational Deep Embedding: A Systematic Architecture Search with Multi-Quadrant Evaluation

  cs.LG · 2026-04-29 · 3 theorem links

  Conv-VaDE with four-dimensional architecture search yields interpretable EEG microstates via generative decoding, achieving peak GEV of 0.730 and silhouette 0.229 at K=4 with depth-4 networks on ten-participant resting-state data.

• Agents of Chaos

  cs.AI · 2026-02-23 · 3 theorem links

  An exploratory red-teaming study documents eleven cases of security, privacy, and governance failures in autonomous language-model agents with tool access and persistent memory.

• Griffin: Mixing Gated Linear Recurrences with Local Attention for Efficient Language Models

  cs.LG · 2024-02-29 · 3 theorem links

  Griffin hybrid model matches Llama-2 performance while trained on over 6 times fewer tokens and offers lower inference latency with higher throughput.

• Posterior-Calibrated Causal Circuits in Variational Autoencoders: Why Image-Domain Interpretability Fails on Tabular Data

  cs.LG · 2026-03-22 · 3 theorem links

  Tabular VAEs show ~50% lower causal circuit modularity than image VAEs, with beta-VAE CES collapsing to 0.043 versus 0.133 due to reconstruction degradation, challenging direct transfer of image interpretability techniques.

• Measuring short-form factuality in large language models

  cs.CL · 2024-11-07 · 3 theorem links

  SimpleQA is a new benchmark of short, single-answer factual questions collected adversarially against GPT-4 to evaluate LLM factuality and confidence calibration.

• One Step Diffusion via Shortcut Models

  cs.LG · 2024-10-16 · 3 theorem links

  Shortcut models enable high-quality single or few-step sampling in diffusion models with one network and training phase by conditioning on desired step size.

• Contraction and Hourglass Persistence for Learning on Graphs, Simplices, and Cells

  cs.LG · 2026-04-19 · 3 theorem links

  Hourglass Persistence interleaves sequences of graph inclusions and contractions to produce more expressive topological features than standard persistent homology for learning on graphs and higher-order complexes.

• Epistemic Observability in Language Models

  cs.DC · 2026-03-20 · 3 theorem links

  Language models cannot be made epistemically honest under text-only supervision, but per-token entropy provides a detectable signal with AUC 0.757 that generalizes across model families.

• Overcoming Dynamics-Blindness: Training-Free Pace-and-Path Correction for VLA Models

  cs.RO · 2026-05-12 · 3 theorem links

  Pace-and-Path Correction decomposes a quadratic cost minimization into orthogonal pace and path channels to correct chunked actions in VLA models, raising success rates by up to 28.8% in dynamic settings.

• Quantitative Linear Logic

  cs.LO · 2026-05-13 · 3 theorem links

  pQLL calculi assign real-valued strength to proofs, generalize hypersequent and deep inference systems, prove cut elimination, and achieve completeness for soft residuated lattices, recovering MALL as p goes to infinity.

• Frequency-Space Mechanics: A Sequence and Coordinate-Free Representation for Protein Function Prediction

  q-bio.BM · 2026-05-12 · 3 theorem links

  Vibrational mode graphs from molecular dynamics enable sequence-free protein function prediction via graph neural networks, with entrainment improving signals for collective dynamics.

• PolitNuggets: Benchmarking Agentic Discovery of Long-Tail Political Facts

  cs.AI · 2026-05-13 · 3 theorem links

  PolitNuggets is a multilingual benchmark showing that AI agents struggle with fine-grained accuracy and efficiency when discovering long-tail political facts for elite biographies, linking performance to short-context extraction, multilingual robustness, and tool use.

• MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGI

  cs.CL · 2023-11-27 · 3 theorem links

  MMMU provides 11.5K heterogeneous college-level multimodal questions that current models solve at 56-59% accuracy, establishing a new standard for expert multimodal evaluation.

• AgentCoder: Multi-Agent-based Code Generation with Iterative Testing and Optimisation

  cs.CL · 2023-12-20 · 3 theorem links

  A three-agent loop of code generation, test creation, and execution feedback lifts pass@1 to 96.3% on HumanEval and 91.8% on MBPP for GPT-4 while using roughly half the tokens of prior state-of-the-art.

• Audio Flamingo 3: Advancing Audio Intelligence with Fully Open Large Audio Language Models

  cs.SD · 2025-07-10 · 3 theorem links

  Audio Flamingo 3 introduces an open large audio-language model achieving new state-of-the-art results on over 20 audio understanding and reasoning benchmarks using a unified encoder and curriculum training on open data.

• Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Models

  cs.CV · 2023-06-08 · 3 theorem links

  Video-ChatGPT is a multimodal model that combines a video visual encoder with an LLM to understand and generate conversations about videos, trained on a new dataset of 100,000 video-instruction pairs.

• Image Restoration via Diffusion Models with Dynamic Resolution

  cs.CV · 2026-05-14 · 3 theorem links

  Dynamic resolution priors enable faster diffusion-based image restoration by operating in lower-dimensional subspaces, with adapted methods outperforming prior DM approaches on most tasks.

• Non-crystallographic systems of integers over composition algebras

  math.RA · 2026-05-14 · 3 theorem links

  A new free rank-8 Z[φ]-order is built by doubling the icosian ring, equipped with a 240-element H4⊕H4 finite root shell, shown to be self-dual under the polar norm and free of octonion-stable nonzero isotropic gluings in its first trace-integral discriminant tower.

• Convergence of difference inclusions via a diameter criterion

  math.OC · 2026-05-14 · 3 theorem links

  A diameter criterion tied to a potential function certifies convergence of difference inclusions, enabling discrete proofs for first-order optimization methods with diminishing steps.

• Branch-width of represented matroids in matrix multiplication time

  cs.DS · 2026-05-14 · 3 theorem links

  O(n² + n^ω)-time algorithm decides if branch-width of a matrix-represented matroid over a finite field is at most k.

• GUI-R1 : A Generalist R1-Style Vision-Language Action Model For GUI Agents

  cs.CV · 2025-04-14 · 3 theorem links

  GUI-R1 uses reinforcement fine-tuning with GRPO on a small curated dataset to create a generalist vision-language action model that outperforms prior GUI agent methods across mobile, desktop, and web benchmarks using only 0.02% of the data.

• Matrix-Space Reinforcement Learning for Reusing Local Transition Geometry

  cs.LG · 2026-05-14 · 3 theorem links

  MSRL represents trajectory segments as PSD matrices to prove additive composition properties and bootstrap value functions for better transfer, reaching 0.73 AUC versus 0.57-0.65 baselines.

• A Study on Type-2 Isomorphic Circulant Graphs: Part 8: $C_{432}(R)$, $C_{6750}(S)$ -- each has 2 types of Type-2 isomorphic circulant graphs

  math.CO · 2026-05-14 · 3 theorem links

  Two families of circulant graphs C_432(R) and C_6750(S) each possess Type-2 isomorphic variants for two values of m.

• 3D Diffusion Policy: Generalizable Visuomotor Policy Learning via Simple 3D Representations

  cs.RO · 2024-03-06 · 3 theorem links

  DP3 uses compact 3D representations from sparse point clouds inside diffusion policies to learn generalizable visuomotor skills from few demonstrations, reporting 24% gains in simulation and 85% success on real robots.

next page →

← previous page