Every paper checked — nothing hidden

All papers

Every paper checked for brief #4 of 12–19 Oct 2026 — nothing is hidden. Brief papers first.

All17,313 In brief24 More (on-topic)3,333 Not relevant13,754 Not assessed87 Older352 Could not confirm93 Blogs with no paper266 All blogs284

Paper	Topic	Relevance
FETALFusion: Resolution-Aware State-Space Modeling for Multi-Domain Fetal Head Segmentation FETALFusion combines CNN and Mamba architectures to segment fetal heads accurately across different ultrasound acquisition methods. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Presents a medical segmentation model, not LLM agents.
FETALFusion: Resolution-Aware State-Space Modeling for Multi-Domain Fetal Head Segmentation FETALFusion combines CNN and Mamba architectures to segment fetal heads accurately across different ultrasound acquisition methods. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Focuses on medical image segmentation, not language model evaluation.
FETALFusion: Resolution-Aware State-Space Modeling for Multi-Domain Fetal Head Segmentation FETALFusion combines CNN and Mamba architectures to segment fetal heads accurately across different ultrasound acquisition methods. Preprint	Computer vision for industrial applications via arxiv · 15 Oct, 06:17	Not relevant Addresses fetal ultrasound segmentation, not industrial manufacturing inspection.
FETALFusion: Resolution-Aware State-Space Modeling for Multi-Domain Fetal Head Segmentation FETALFusion combines CNN and Mamba architectures to segment fetal heads accurately across different ultrasound acquisition methods. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Focuses on ultrasound image segmentation, not multimodal vision-language models.
FETALFusion: Resolution-Aware State-Space Modeling for Multi-Domain Fetal Head Segmentation FETALFusion combines CNN and Mamba architectures to segment fetal heads accurately across different ultrasound acquisition methods. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Addresses medical imaging, not AI deployment safety or guardrails.
Virtual Head Attention Virtual Head Attention uses two linear operations to compress key-value cache overhead without increasing training compute or sacrificing output quality. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Focuses on attention architecture for KV cache, not retrieval-augmented generation.
Virtual Head Attention Virtual Head Attention uses two linear operations to compress key-value cache overhead without increasing training compute or sacrificing output quality. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Proposes an attention mechanism, not agent architectures or planning.
Virtual Head Attention Virtual Head Attention uses two linear operations to compress key-value cache overhead without increasing training compute or sacrificing output quality. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Presents an architectural design, not an LLM benchmark.
Virtual Head Attention Virtual Head Attention uses two linear operations to compress key-value cache overhead without increasing training compute or sacrificing output quality. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Addresses core transformer attention mechanics, not software engineering.
Virtual Head Attention Virtual Head Attention uses two linear operations to compress key-value cache overhead without increasing training compute or sacrificing output quality. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on attention efficiency, not AI safety or guardrails.
Virtual Head Attention Virtual Head Attention uses two linear operations to compress key-value cache overhead without increasing training compute or sacrificing output quality. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Concerns general transformer attention, not multimodal-specific processing.
Better Capturing Interactions Between Products in~Retail: Revisited Negative Sampling for~Basket Choice Modeling The paper proposes a shopping basket choice model using specialized negative sampling to capture product interactions from retail transactions. Peer-reviewed	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Focuses on retail recommendation embeddings, not small language models or inference.
Better Capturing Interactions Between Products in~Retail: Revisited Negative Sampling for~Basket Choice Modeling The paper proposes a shopping basket choice model using specialized negative sampling to capture product interactions from retail transactions. Peer-reviewed	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on retail basket choice modeling, not LLM agents.
Better Capturing Interactions Between Products in~Retail: Revisited Negative Sampling for~Basket Choice Modeling The paper proposes a shopping basket choice model using specialized negative sampling to capture product interactions from retail transactions. Peer-reviewed	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Addresses product interaction modeling, not AI deployment safety.
Better Capturing Interactions Between Products in~Retail: Revisited Negative Sampling for~Basket Choice Modeling The paper proposes a shopping basket choice model using specialized negative sampling to capture product interactions from retail transactions. Peer-reviewed	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Concerns basket choice embeddings, not LLM evaluation.
BELT : BEhavior-Aligned Low-Rank KV Cache Compression with Learned Top-k Rank Allocation BELT compresses key-value caches using attention-distribution alignment and learned global rank allocation across model heads. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Presents a compression framework, not an evaluation benchmark.
BELT : BEhavior-Aligned Low-Rank KV Cache Compression with Learned Top-k Rank Allocation BELT compresses key-value caches using attention-distribution alignment and learned global rank allocation across model heads. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on KV cache reduction, not multi-step agent planning.
BELT : BEhavior-Aligned Low-Rank KV Cache Compression with Learned Top-k Rank Allocation BELT compresses key-value caches using attention-distribution alignment and learned global rank allocation across model heads. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Concerns cache optimization rather than safety, jailbreaks, or hallucination.
Seeing Depth but Choosing Cues in Vision-Language Models The paper analyzes how vision-language models favor 2D image cues over internal 3D depth representations when judging relative distances. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Studies perception in VLMs, not agent planning or tool use.
Seeing Depth but Choosing Cues in Vision-Language Models The paper analyzes how vision-language models favor 2D image cues over internal 3D depth representations when judging relative distances. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Analyzes depth cues and internal representations, not benchmark methodology.
Seeing Depth but Choosing Cues in Vision-Language Models The paper analyzes how vision-language models favor 2D image cues over internal 3D depth representations when judging relative distances. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Investigates visual depth perception heuristics, not deployment safety or guardrails.
Seeing Through the Forgery Is Not Enough: Regressive Sycophancy under Fabricated Evidence in Medical Vision-Language Models The authors show that medical vision-language models frequently abandon correct clinical answers when prompted with fabricated medical documents. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Concerns sycophancy evaluation, not model compression or efficiency.
Seeing Through the Forgery Is Not Enough: Regressive Sycophancy under Fabricated Evidence in Medical Vision-Language Models The authors show that medical vision-language models frequently abandon correct clinical answers when prompted with fabricated medical documents. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Focuses on sycophantic alignment to fake evidence, not RAG systems.
Seeing Through the Forgery Is Not Enough: Regressive Sycophancy under Fabricated Evidence in Medical Vision-Language Models The authors show that medical vision-language models frequently abandon correct clinical answers when prompted with fabricated medical documents. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Investigates VLM sycophancy, not tool-using or planning agents.
Seeing Through the Forgery Is Not Enough: Regressive Sycophancy under Fabricated Evidence in Medical Vision-Language Models The authors show that medical vision-language models frequently abandon correct clinical answers when prompted with fabricated medical documents. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Focuses on medical vision-language models, not code LLMs.
Saliency-Guided Replay for Continual Learning in Deep Spiking Neural Networks The paper presents a saliency-guided patch replay method to mitigate catastrophic forgetting in spiking neural networks under tight memory budgets. Preprint	Computer vision for industrial applications via arxiv · 15 Oct, 06:17	Not relevant Evaluated on CIFAR datasets, not industrial defect detection or manufacturing.
Saliency-Guided Replay for Continual Learning in Deep Spiking Neural Networks The paper presents a saliency-guided patch replay method to mitigate catastrophic forgetting in spiking neural networks under tight memory budgets. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on continual learning in spiking neural networks, not agents.
Saliency-Guided Replay for Continual Learning in Deep Spiking Neural Networks The paper presents a saliency-guided patch replay method to mitigate catastrophic forgetting in spiking neural networks under tight memory budgets. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Concerns continual learning replay, not AI deployment safety or robustness.
Saliency-Guided Replay for Continual Learning in Deep Spiking Neural Networks The paper presents a saliency-guided patch replay method to mitigate catastrophic forgetting in spiking neural networks under tight memory budgets. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes a replay strategy rather than LLM benchmarking.
Saliency-Guided Replay for Continual Learning in Deep Spiking Neural Networks The paper presents a saliency-guided patch replay method to mitigate catastrophic forgetting in spiking neural networks under tight memory budgets. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Deals with neuromorphic vision networks, not multimodal foundation models.
Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models The paper proposes predictability-aware supervised fine-tuning to resolve training and inference order mismatches in diffusion language models. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Tests on code benchmarks, but the core contribution is general DLM fine-tuning.
Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models The paper proposes predictability-aware supervised fine-tuning to resolve training and inference order mismatches in diffusion language models. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Focuses on fine-tuning diffusion language models, not document retrieval.
Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models The paper proposes predictability-aware supervised fine-tuning to resolve training and inference order mismatches in diffusion language models. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Focuses on diffusion model training dynamics, not model compression or serving.
Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models The paper proposes predictability-aware supervised fine-tuning to resolve training and inference order mismatches in diffusion language models. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes a training method, not a benchmark or evaluation framework.
Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models The paper proposes predictability-aware supervised fine-tuning to resolve training and inference order mismatches in diffusion language models. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on diffusion model SFT, not deployment safety or guardrails.
Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models The paper proposes predictability-aware supervised fine-tuning to resolve training and inference order mismatches in diffusion language models. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Deals with diffusion language models, not multimodal vision-language models.
Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models The paper proposes predictability-aware supervised fine-tuning to resolve training and inference order mismatches in diffusion language models. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Concerns training diffusion language models, not autonomous planning agents.
SWEPatchTest: Evaluating Software Tests Beyond Reproduction SWEPatchTest provides a benchmark to evaluate whether agent-generated test suites correctly accept valid patches and reject invalid ones. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on evaluating generated software test suites, not general agent frameworks.
SWEPatchTest: Evaluating Software Tests Beyond Reproduction SWEPatchTest provides a benchmark to evaluate whether agent-generated test suites correctly accept valid patches and reject invalid ones. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on test-suite benchmarking rather than deployment safety or jailbreaks.
Cost-Efficient Privacy Enforcement for Email-Generating LLM Agents under Strategic Information Requests The paper introduces local contract-conditioned rewriting to prevent sensitive data leakage in email-generating language model agents cost-effectively. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes an enforcement framework rather than general LLM evaluation benchmarks.
Correcting or Continuing? What On-Policy Distillation Learns from Student Errors The study shows that training on-policy distillation exclusively on student errors degrades accuracy rather than improving it. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Analyzes student error distillation rather than establishing evaluation benchmarks.
Correcting or Continuing? What On-Policy Distillation Learns from Student Errors The study shows that training on-policy distillation exclusively on student errors degrades accuracy rather than improving it. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Examines training distillation dynamics, not deployment safety or guardrails.
Correcting or Continuing? What On-Policy Distillation Learns from Student Errors The study shows that training on-policy distillation exclusively on student errors degrades accuracy rather than improving it. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Investigates distillation training dynamics, not autonomous LLM agents.
LatentRemedy: Efficient Remedy For Non-Prefix KV Cache Reuse in Multi-Agent Systems LatentRemedy uses task-conditioned latent tokens to allow safe key-value cache reuse across conflicting tasks in multi-agent systems. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes an architectural inference remedy, not an evaluation benchmark.
LatentRemedy: Efficient Remedy For Non-Prefix KV Cache Reuse in Multi-Agent Systems LatentRemedy uses task-conditioned latent tokens to allow safe key-value cache reuse across conflicting tasks in multi-agent systems. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on inference cache efficiency, not deployment safety or jailbreaks.
LatentRemedy: Efficient Remedy For Non-Prefix KV Cache Reuse in Multi-Agent Systems LatentRemedy uses task-conditioned latent tokens to allow safe key-value cache reuse across conflicting tasks in multi-agent systems. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Focuses on KV cache reuse across agents, not retrieval-augmented generation.
LatentRemedy: Efficient Remedy For Non-Prefix KV Cache Reuse in Multi-Agent Systems LatentRemedy uses task-conditioned latent tokens to allow safe key-value cache reuse across conflicting tasks in multi-agent systems. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Not focused on code generation or software engineering tasks.
LatentRemedy: Efficient Remedy For Non-Prefix KV Cache Reuse in Multi-Agent Systems LatentRemedy uses task-conditioned latent tokens to allow safe key-value cache reuse across conflicting tasks in multi-agent systems. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Concerns text KV cache reuse, not multimodal models.
SHELDON: Reasoning through Dual Search over Hypothesis and Observation Spaces SHELDON introduces a dual-search reasoning framework that independently explores hypotheses and observation spaces to avoid confirmation bias. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes a reasoning architecture rather than a benchmark or evaluation framework.
SHELDON: Reasoning through Dual Search over Hypothesis and Observation Spaces SHELDON introduces a dual-search reasoning framework that independently explores hypotheses and observation spaces to avoid confirmation bias. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Studies test-time search and reasoning, not model compression or efficiency.
SHELDON: Reasoning through Dual Search over Hypothesis and Observation Spaces SHELDON introduces a dual-search reasoning framework that independently explores hypotheses and observation spaces to avoid confirmation bias. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on scientific discovery search, not deployment safety or guardrails.
ChampionKV: Head-Wise Adaptive Budget Allocation for KV Cache Compression ChampionKV adaptively allocates key-value cache memory budgets across attention heads using observation-window concentration proxies. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Develops a cache compression method rather than evaluation methodology.
ChampionKV: Head-Wise Adaptive Budget Allocation for KV Cache Compression ChampionKV adaptively allocates key-value cache memory budgets across attention heads using observation-window concentration proxies. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on KV cache compression, not multi-step agent planning.
ChampionKV: Head-Wise Adaptive Budget Allocation for KV Cache Compression ChampionKV adaptively allocates key-value cache memory budgets across attention heads using observation-window concentration proxies. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Deals with inference compression rather than safety, jailbreaks, or hallucination.
Can Machine Learning Potentials Be More Accurate Than the Methods They Are Trained On? The authors propose methods to train machine learning potentials that transfer across quantum-chemical methods to exceed single-method training accuracy. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Focuses on chemical simulation potentials rather than LLM benchmarks.
Can Machine Learning Potentials Be More Accurate Than the Methods They Are Trained On? The authors propose methods to train machine learning potentials that transfer across quantum-chemical methods to exceed single-method training accuracy. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant About quantum chemistry machine learning potentials, not language model agents.
Can Machine Learning Potentials Be More Accurate Than the Methods They Are Trained On? The authors propose methods to train machine learning potentials that transfer across quantum-chemical methods to exceed single-method training accuracy. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Concerns molecular physics models, not small language models or inference.
Can Machine Learning Potentials Be More Accurate Than the Methods They Are Trained On? The authors propose methods to train machine learning potentials that transfer across quantum-chemical methods to exceed single-method training accuracy. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Addresses quantum simulation accuracy, not AI deployment safety or guardrails.
Scaling Vision Transformers Without Training: Prune-and-Merge Token Compression The paper introduces a training-free token pruning and merging framework with new robustness metrics for vision transformers. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on token reduction efficiency rather than safety, guardrails, or jailbreaks.
Scaling Vision Transformers Without Training: Prune-and-Merge Token Compression The paper introduces a training-free token pruning and merging framework with new robustness metrics for vision transformers. Preprint	Computer vision for industrial applications via arxiv · 15 Oct, 06:17	Not relevant Applies token reduction to standard vision models, not industrial inspection.

10,261–10,320 of 13,754← Previous Next →