Every paper checked — nothing hidden
All papers
Every paper checked in the week of 12–18 Oct 2026 · LLM evaluation and benchmarks — nothing is hidden. Brief papers first.
All3,345In brief3More (on-topic)587Not relevant2,721Not assessed24Older45Could not confirm25Blogs with no paper54All blogs57
| Paper | Topic | Relevance | Feedback |
|---|---|---|---|
| ML-CLAPPS: A Probabilistic Framework for Imbalanced Multilabel Learning under Class Prioritization This paper introduces a learnable probabilistic instance sampler to handle severe class imbalance and prioritize critical diagnoses in multilabel learning. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes an imbalanced training sampler rather than an evaluation benchmark. | |
| Mobility Helps Learning: Unsupervised Model Adaptation for Object Recognition via Movement This paper presents an unsupervised test-time adaptation method that uses movement-induced prediction variance to adapt object recognition models in mobile agents. Peer-reviewed | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Introduces an unsupervised test-time adaptation technique, not an evaluation framework. | |
| Frequency-Step Cache: Separating Synchronization and Repair for Efficient Autoregressive Video Diffusion This paper introduces a training-free caching method that separates cache synchronization and computation repair to accelerate autoregressive video diffusion generation. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Introduces a caching algorithm rather than an evaluation benchmark. | |
| Representational Difference Distillation Proposes a knowledge distillation method that focuses learning signals on specific structural disagreements between teacher and student representations. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes a distillation loss formulation rather than benchmarking methodologies. | |
| Regime-Aware Router Learning for Multi-Domain Medical Image Generation with Diffusion Experts Evaluates router-learning strategies for combining domain-specialized diffusion experts in multi-domain medical image generation. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Presents a routing framework for diffusion experts, not model evaluation methods. | |
| Elephant in the Fridge: Constant-Memory Frame Packing for Long Video Understanding Proposes a training-free frame packing method enabling vision-language models to process long videos with constant memory consumption. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes a video compression technique rather than benchmark or evaluation methodology. | |
| Learning $ω$-Powered Conditional Distributions with Importance-Weighted Flow Matching The paper introduces an importance-weighted flow matching approach to explicitly learn powered conditional distributions without standard classifier-free guidance. Peer-reviewed | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on generative flow algorithms rather than LLM benchmarking. | |
| KARAMEL: Knowledge-Aware Ranking for Multimodal Entity Linking in the Arts via Heterogeneous Graphs The paper introduces a multimodal entity linking method that combines vision-language models with heterogeneous graph networks for cultural heritage data. Peer-reviewed | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Not about language model benchmark methodologies or LLM evaluation. | |
| Variational Filtering for Bayesian Last Layers with an Independent Binary Probit Surrogate The paper proposes a closed-form variational filtering algorithm for the last layer of deep classifiers to improve uncertainty calibration. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Not about LLM evaluation or benchmark methodology. | |
| Code-Interactive Visual Reasoning for Financial Candlestick Understanding The paper introduces a model and dataset for financial candlestick reasoning by generating and executing code to gather visual evidence. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Introduces a financial reasoning benchmark, but primary focus is multimodal code-interactive chart reasoning. | |
| JtT: Rethinking Point Trajectory Diffusion for Diverse and Rigid Motion from an Image The paper presents a diffusion transformer operating on per-frame velocities to generate rigid point trajectories from a single image. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on point trajectory diffusion generation, not LLM benchmark evaluation. | |
| Carve VFM: Carving Specialists out of Generalists, Automatic Task-Specific Compression of Vision Foundation Models The paper proposes a structured pruning method to compress vision foundation models into compact task-specific models under parameter budgets. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Introduces a pruning technique, not LLM benchmarks or evaluation methods. | |
| Barycentric Guidance: Turning Foundation Image Editors into Continuous Affective Controllers The paper proposes a training-free guidance method enabling continuous valence-arousal emotional control in foundation image editors. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes an image editing controller, not LLM evaluation benchmarks. | |
| HybridDualPathNet: Cross-Fusion Attention of EfficientNetV2 and Swin Transformer for Brain MRI Classification of Alzheimer's and Parkinson's Disease The authors introduce a hybrid architecture combining EfficientNetV2 and Swin Transformer to classify Alzheimer's and Parkinson's disease from brain MRI. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Evaluates medical classification models, not language model benchmarks. | |
| Efficient Fréchet Distributional Decoder Alignment for Latent Generative Models Proposes a post-training decoder alignment method for latent generative models to resolve distribution mismatches between training and generation latents. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Addresses generative model post-training alignment rather than LLM benchmarking. | |
| CÆNL: Collapse-Based Active Neural Learning with Objective-Aligned Representation Control Investigates active learning dynamics by regularizing representation collapse and managing sample acquisition across training iterations. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on active learning algorithms rather than LLM evaluation benchmarks. | |
| Calibrated Disambiguation for Partial Multi-label Learning Proposes a calibrated disambiguation model to optimize sample weighting when training classifiers on datasets with noisy or excessive labels. Peer-reviewed | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Deals with multi-label learning, not language model evaluation or benchmarks. | |
| Deficit-Driven Stratified Acquisition for Building STRATA-S12: An Open SAR-Optical Corpus for Self-Supervised Learning Introduces a deficit-driven stratified acquisition algorithm to create a balanced, co-registered satellite imagery corpus for self-supervised Earth observation. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Presents an acquisition strategy for satellite imagery, not an evaluation framework for LLMs. | |
| A Differentiable 3D Scene Graph Metric via Contextual Hellinger Triplet Geometry Proposes a differentiable graph-level loss function to measure semantic and structural discrepancies between predicted 3D scene graphs and ground truth. Peer-reviewed | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes a training loss for 3D scene graphs, not LLM benchmark evaluation methods. | |
| DT-LoRA: Reaching Beyond the Tangent Space of Low-Rank Adaptation Introduces a low-rank adaptation method that splits rank budgets into stable and dynamic branches to capture residual gradients outside the tangent space. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Presents a parameter-efficient fine-tuning algorithm rather than an LLM evaluation benchmark or framework. | |
| Beyond the End-Effector: Learning Whole-Arm Contact Manipulation from Human Demonstrations This paper proposes a hierarchical framework for learning whole-arm robot contact manipulation tasks from human demonstrations. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Studies robotic arm contact trajectory optimization, not LLM evaluation or benchmark trustworthiness. | |
| SPARCL: Spatially Adaptive Regional Coverage for Active Learning in Chest X-ray Analysis This paper introduces an active learning framework balancing uncertainty with submodular spatial coverage for chest X-ray lesion segmentation. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes an active learning acquisition method and medical consistency metrics, not an LLM benchmark. | |
| Sea-Net: A Small and Inherently Interpretable Neural Network for Time Series Classification This paper presents a lightweight, interpretable neural network architecture for resource-constrained time-series classification on microcontrollers. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on time-series classifiers, not LLM evaluation and benchmarking methods. | |
| SYNPRED: A Synergistic Approach to Multimodal Learning for Clinical Prediction This paper proposes a multimodal variational autoencoder framework that separates task-relevant information to improve clinical predictions even with missing modalities. Peer-reviewed | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Evaluates medical VAEs, not language model benchmark methodologies. | |
| Robust Logical Anomaly Detection under Capture Shifts via Synthetic Anomalies Examines industrial logical anomaly detection under capture shifts and proposes a reference-free classifier trained on synthetic anomalies. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Evaluates visual anomaly detection methods rather than language model benchmarks or judges. | |
| Simplifying and Improving On-Policy Distillation for Few-Step Diffusion Proposes Simplified Matching Distillation, a regression-based on-policy method that enables four-step generation for diffusion models. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Evaluates generative diffusion models, not LLM benchmarks or evaluation methods. | |
| Beyond the Encoder: Object-Centric Representations Emerging from Segmentation Foundation Models This paper evaluates whether object pointers from the Segment Anything Model can transfer to downstream clinical tumor prediction tasks. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Analyzes feature transferability in medical imaging, not LLM evaluation. | |
| Unbalanced W-Flow: Unbalanced Wasserstein Drifting under Finite-Batch Marginal Fluctuations The paper proposes an unbalanced optimal transport method to reduce minibatch fluctuations in teacher-free one-step generative models. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Covers generative transport theory, not LLM evaluation or benchmarks. | |
| MajorTOM:Elliot Pretrain, A Multi-Sensor, Multi-Temporal Pretraining Dataset for Earth Observation This paper introduces a multi-sensor, multi-temporal satellite imagery dataset designed for pretraining Earth observation foundation models. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Presents an Earth observation dataset, not an LLM benchmark. | |
| Grid or Nodes? Probing Object Memory in World Models Under Partial Observability The paper compares spatial feature grids and tracked node coordinates for maintaining object memory under partial robot observability. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on tracking memory in robotics, not LLM benchmarks. | |
| Texture Interpolation as Texton Matching This paper studies texture interpolation using appearance-aware texton matching instead of standard positional coordinate blending. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on image texture synthesis, not LLM evaluation. | |
| AIRRI: Towards a Human-Readable, Machine-Resistant Reading Interface This paper presents an interface framework applying visual transformations and adversarial attacks to create assignment text resistant to machine extraction. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Evaluates OCR degradation, not general LLM benchmarking methods. | |
| Registration is a Representation Problem: Solving Whole-Body PET/CT Registration This paper presents an anatomical representation approach driving iterative solvers for longitudinal whole-body PET and CT image registration. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Addresses medical image registration, not LLM evaluation or benchmarks. | |
| Scaling Online Optimal Transport Minibatches for Flow Matching This paper develops an online entropic optimal transport approximation to scale minibatch coupling for stable flow matching. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on generative flow optimization, not LLM evaluation benchmarks. | |
| Cepheid: Decoupling Scale from Structure for Cross-Family Motion Completion This paper proposes a dimensionless prior to complete dense mesh motion from sparse observations across different geometry families. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on mesh motion estimation rather than LLM benchmarking methods. | |
| ByteMark: A Unified Byte-Level Watermark for Detection Across Language Models This paper presents a byte-level watermarking method that allows a single detector to verify text across models with different tokenizers. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Presents a watermarking method rather than LLM benchmarks or evaluation methods. | |
| KedgeFlow: A Programmable Control Plane for Enterprise AI Agents The paper proposes a control plane enforcing structured contracts and single-use allowances to prevent unauthorized actions by enterprise AI agents. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Presents a control plane architecture, not an evaluation benchmark. | |
| SEEK: Agentic Knowledge Graph Retrieval for Secure Code Generation The paper introduces an agentic knowledge graph retrieval framework that uses vulnerability taxonomies to generate secure code. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Presents a generation framework rather than a general evaluation benchmark. | |
| Only What You've Signed up For: Secure Workflows with Constrained Diffusion Agents The paper presents a framework enforcing security policies during the denoising steps of diffusion language models to prevent prompt injection. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on a security defense mechanism, not LLM evaluation benchmarks. | |
| From Acoustic Sensing to Physical Actions: A Comprehensive Survey of Security Risks in Edge-Deployed Audio LLM-Based Agents The paper surveys security vulnerabilities and defenses in edge-deployed audio language model agents across the sensing-to-action pipeline. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Survey of audio agent security, not an evaluation benchmark. | |
| Poisoned into Action: Context Attacks on Typed Decision Workflows The paper studies how context poisoning triggers unauthorized state changes in typed decision workflows and evaluates an execution-aware defense protocol. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Designs an attack evaluation protocol for workflows, not general LLM benchmarking. | |
| Robust Detection of Directional Adversarial Attacks in Deep Neural Networks for Radiological Imaging The paper presents a framework using random noise and Sigmoid function analysis to detect directional adversarial attacks on medical imaging models. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes an adversarial detector for vision, not an LLM benchmark. | |
| To Cache or Not to Cache: Understanding and Mitigating Safety Incompatibility in Cross-Model KV Cache Reuse The paper studies safety behavioral deviations caused by cross-model key-value cache reuse and proposes an alignment technique to mitigate them. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Not introducing a general benchmark or evaluation methodology for LLMs. | |
| Can Rényi Accountants Beat Lossless Accountants for DP-SGD? This paper demonstrates that Renyi differential privacy accountants can provide tighter privacy bounds for DP-SGD than lossless accountants. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Theoretical differential privacy analysis, not an LLM benchmark. | |
| Tight End-to-End Accounting for Differentially Private Synthetic Data This paper develops accounting techniques to maintain tight differential privacy guarantees across complex synthetic data generation pipelines. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Theoretical privacy accounting framework, not an LLM benchmark. | |
| SSM-Attack: Learning Where to Perturb for Black-Box Face Privacy This paper proposes using state space models to guide black-box image perturbations that disrupt facial recognition systems. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Introduces an adversarial attack method rather than an evaluation benchmark. | |
| COMPASS: Membership Inference Attack Against Vision Large Language Models via Adaptive Noise Calibration This paper investigates membership inference attacks on vision-language models using low-noise Gaussian perturbations to detect training data exposure. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on privacy attacks on multimodal models rather than benchmark evaluation. | |
| Privacy-Preserving GNN Inference on Financial Transaction Graphs via Homomorphic Encryption This paper proposes an edge-centric graph neural network inference framework using homomorphic encryption for collaborative multi-bank transaction analysis. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Presents encrypted GNN inference for AML, not LLM evaluation or benchmarking. | |
| Kairos: Transactional Security Enforcement at the System Boundary for LLM Agents This paper introduces a transactional isolation monitor that delays irreversible actions to stop indirect prompt injections in software agents. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes a system boundary monitor for agents, not an LLM benchmark. | |
| Towards Safe Autonomy for Computer Use This paper proposes a dual-LLM architecture to establish trust boundaries and prevent indirect prompt injections in visual computer-use agents. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes a security architecture against prompt injection, not an LLM benchmark. | |
| Autonomous Cyber Operations Gym for Training and Evaluating AI Models for Defence The paper introduces a simulation gym for training and evaluating autonomous cyber defence models and network intrusion detectors. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Proposes an emulation gym for cyber operations, not an LLM evaluation benchmark. | |
| Fin_Agent at FinEval 2026: Task-Aware Methods for Multilingual Financial NLP The paper describes a system submission combining rules and language models for multilingual financial named-entity recognition and question answering. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant A shared-task competition entry, not an evaluation methodology or benchmark study. | |
| Does System One Transfer? Calibrated Typed-Decision Models, Label Budgets, and Translation for Indonesian Regional Languages The paper evaluates the cross-lingual transfer, calibration, and translation sensitivity of fast typed-decision models on regional Indonesian languages. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on cross-lingual evaluation of classifiers rather than general LLM benchmark analysis. | |
| The AI's Poker Face: Contrasting Persuasion and Sycophancy in Voice-Based Multi-Turn Conversations The paper studies opinion change, rhetorical strategies, and sycophancy in spoken multi-turn dialogues between humans and language models. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Investigates human belief change via user study rather than LLM benchmarking methods. | |
| AutoEnvScaling: Automating the Data Flywheel with Terminal Agents The paper presents AutoEnvScaling, using terminal agents to automatically generate reinforcement learning environments from model rollouts. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Focuses on environment generation and training methods rather than LLM benchmark evaluation. | |
| Simple Linear Corrections for Compression in Late Interaction The paper proposes a linear score correction method to recover retrieval quality in compressed late-interaction document representations. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Evaluates retrieval methods on benchmarks but does not introduce general LLM evaluations. | |
| Modeling the Arrival Time of Information-Bearing Events in Natural Language The paper uses neural temporal point processes to model the timing of information-bearing events in human and model generated text. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Analyzes token arrival statistics, not an evaluation benchmark. | |
| ASSET: Adaptive Speech Segmentation via Reinforcement Learning The paper introduces ASSET, a reinforcement learning method that adaptively compresses speech encoder outputs into continuous audio tokens. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Presents an audio compression method, not an evaluation benchmark. | |
| Multimodal Co-Generation With Continuous Bitstream Diffusion The paper applies continuous bitstream diffusion to joint and conditional generation across multimodal data without learned tokenizers. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Introduces a generative model framework, not an evaluation benchmark. | |
| Tokenization-Invariant Watermarking: A Byte-Local Method for LLM Text The paper presents a byte-local text watermarking method that scores outputs without invoking tokenizers during detection. Preprint | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | Not relevant Introduces a watermarking method rather than an evaluation benchmark. |
1,921–1,980 of 3,345← PreviousNext →