Search: “llm”

Measuring Cross-Task Behavioral Consistency in Language Model Agents

arXiv:2608.13598v1 Announce Type: new Abstract: Agent evaluation relies almost entirely on outcome metrics such as success rate, which capture whether an agent succeeds but not how consistently it beh

17 Aug 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Never the Number: Structural Abstention for AI Systems Whose Answers Are Consumed as Fact

DGX agent

arXiv:2608.13926v1 Announce Type: new Abstract: Large language models have made natural language interfaces to databases (NLIDB) newly credible, but LLM text-to-SQL systems fail in a way that matters

agentsarxiv-cs-ai

17 Aug 2026

ScienceFlow: A long-horizon agent for ML research, scientific discovery and beyond

arXiv:2608.14354v1 Announce Type: new Abstract: Enabling LLM agents to sustain productive, stable, and goal-aligned research over extended horizons is a central challenge for autonomous machine learni

17 Aug 2026

Understanding and Mitigating Over-refusal for Large Language Models via Representation Intervention

arXiv:2511.19009v2 Announce Type: replace-cross Abstract: Large language models (LLMs) demonstrate powerful capabilities across various natural language processing tasks,yet their inherent safety vuln

safetyarxiv-cs-cl

17 Aug 2026

Based on an accelerating frontier -> local trajectory, expect a ~30b param 'Mythos at home' by as soon as Jan 2027 (rationalisation below)

model-releasesr-localllama

Including the rationalisation for the data below - this is a more robust version of an earlier post I did similar to this - explaining below: How I chose the comparisons The basic question I’m trying

16 Aug 2026

AQuA: Recursively Self-Improving Quantitative Trading Research Agents

arXiv:2608.12841v1 Announce Type: cross Abstract: We study recursive self-improvement at the level of quantitative-investment research: whether an autonomous system can use evidence from earlier exper

BoardroomAI: Dependency-Aware Human-Steerable Multi-Agent Deliberation through Evolving Decision Graphs

arXiv:2608.13046v1 Announce Type: new Abstract: Organizational decisions are co-created while evidence, constraints, and human priorities continue to evolve. In conventional transcript-based multi-age

Can Spectral-Clipping Enable Better Learning While Forgetting Less for Low-Rank Adaptation?

arXiv:2608.12332v1 Announce Type: new Abstract: In recent years, low-rank adaptation (LoRA) has emerged as a significant paradigm that freezes pre-trained weights and introduces small, learnable adapt

Decomposition of Evidence, Contradiction, and Fragility in Perturbation Responses

arXiv:2608.12935v1 Announce Type: new Abstract: Perturbation methods explain model decisions by measuring prediction changes under altered inputs, but response magnitude tells us only how much a model

researcharxiv-cs-ai

Exploiting Symbolic Heuristics for the Synthesis of Domain-Specific Temporal Planning Guidance using Reinforcement Learning

arXiv:2505.13372v2 Announce Type: replace Abstract: Recent work investigated the use of Reinforcement Learning (RL) for the synthesis of heuristic guidance to improve the performance of temporal plann

researcharxiv-cs-ai

Into the ORBIT for Time Series: Training Regimes for Foundation Models

arXiv:2608.13262v1 Announce Type: cross Abstract: Time series foundation models (TSFMs) have advanced primarily through architectural innovation, while training regimes for large-scale heterogeneous c

safetyarxiv-cs-ai

Large Language Models Pass the History Exam But Miss the <<History>>: A Polish High School Exit Exam Matura Benchmark

model-releasesarxiv-cs-cl

arXiv:2608.12343v1 Announce Type: new Abstract: AI chatbots are widely used by students as knowledge sources, yet LLM benchmarks rarely assess interpretative historical reasoning. We evaluate eight le

LoKiFormer: Locality-aware Attention with Decoupled Knowledge Memory for Efficient Large Language Model Pretraining

arXiv:2608.12419v1 Announce Type: new Abstract: Large language models (LLMs) have achieved remarkable breakthroughs across various applications. However, their architectures remain inefficient in pret

local-aiarxiv-cs-lg

neurosymbolic world models – exactly what I argued for in 2020 in the Next Decade in AI – for the win.

neurosymbolic world models – *exactly* what I argued for in 2020 in the Next Decade in AI – for the win. Jeremy's excellent work here is a great illustration of a very powerful type of approach: LLM-g

safetygary-marcus--x

Position: The Alignment Community is Unintentionally Building a Censor's Toolkit

arXiv:2608.12346v1 Announce Type: new Abstract: This position paper argues that modern AI alignment methods - originally designed to prevent harmful output - are dual-use technologies that may easily

safetyarxiv-cs-ai

Prompts in the Wild: A Large Analyzed Collection of Transactional Prompts in Code

arXiv:2608.12905v1 Announce Type: new Abstract: The behavior of contemporary generative Large Language Models (LLMs) is directly shaped by prompts, unstructured texts that describe the desired output

Security and Detectability Analysis of Unicode Text Watermarking Methods against Large Language Models

arXiv:2512.13325v2 Announce Type: replace-cross Abstract: Securing digital text is becoming increasingly relevant due to the widespread use of large language models. Individuals' fear of losing contro

Self-Referential Induction Increases Response Instability Relative to Unresolvable and Verifiable Questions in Large Language Models

arXiv:2608.13258v1 Announce Type: cross Abstract: Self-referential prompting has been shown to reliably induce large language models to produce first-person reports resembling subjective experience, b

TEMPO: Makespan-Aware Expert-Parallel Load Balancing Across Memory- and Compute-Bound Regimes

arXiv:2608.13057v1 Announce Type: cross Abstract: In expert-parallel (EP) MoE serving, every layer synchronizes at the slowest GPU. Dispatchers balance token counts (EPLB, LPLB, UltraEP) or activated-

The Hidden Evolution of Disguised Visual Context inside the VLM

arXiv:2606.20077v2 Announce Type: replace-cross Abstract: Visual tokens enter Large Language Models (LLMs) as raw, foreign signals. How they are transformed into meaningful representations and interac

researcharxiv-cs-ai

DREAMS: Density Functional Theory Based Research Engine for Agentic Materials Simulation

arXiv:2507.14267v2 Announce Type: replace Abstract: Large language model (LLM) agents can execute long-horizon scientific workflows, but their numerical outputs are difficult to trust: agents lose con

13 Aug 2026

Actionable Hallucination Detection: Translating Latent Uncertainty into Agentic Critique

arXiv:2608.10430v1 Announce Type: cross Abstract: Large Language Models (LLMs) deployed as AI agents frequently exhibit user specification-grounding failures, executing hallucinated, undesired actions

Do Personalized Skills Help Coding Agents? An Empirical Study of Developer Interaction Histories

arXiv:2608.10319v1 Announce Type: cross Abstract: Large language model (LLM)-powered agents have rapidly evolved from code-completion tools into solvers of complex software engineering tasks. As devel

Enhancing Automated Essay Scoring With Three Techniques: Two-Stage Fine-Tuning, Score Alignment, and Self-Training

arXiv:2602.01747v2 Announce Type: replace Abstract: Automated Essay Scoring (AES) plays a crucial role in education by providing scalable and efficient assessment tools. However, in real-world setting

safetyarxiv-cs-cl

How to Dogfood Your AI Chat Agent: A Three-Layer Evaluation Framework with Goal-Directed NPC Simulation

arXiv:2608.09939v1 Announce Type: cross Abstract: Production teams deploying LLM chat agents face a specific quality assurance gap: existing evaluation tools test individual responses or simulate soci

Leveraging Large Language Models for Causal Discovery: a Constraint-based, Argumentation-driven Approach

arXiv:2602.16481v2 Announce Type: replace Abstract: Causal discovery seeks to uncover causal relations from data, typically represented as causal graphs, and is essential for predicting the effects of

safetyarxiv-cs-ai

Mitigating Context Interference for Reliable and Efficient Search Agents

arXiv:2608.10743v1 Announce Type: new Abstract: Recent research empowers Large Language Models (LLMs) as multi-turn search agents to iteratively retrieve and generate outputs until complex tasks are s

agentsarxiv-cs-cl

We’re excited to announce DiG-bench, a new benchmark for discovery! Over the last few weeks we’ve been testing frontier AI models on our nov…

model-releasesemad-mostaque--x

We’re excited to announce DiG-bench, a new benchmark for discovery! Over the last few weeks we’ve been testing frontier AI models on our novel discovery games and seeing how they score. Each game is a

Agentic Router: An Execution-Grounded Continual Learning Approach With Memory

arXiv:2608.09184v1 Announce Type: new Abstract: Large language model (LLM) agents provide a promising interface for command-line-based network operations, but a plausible command may still fail or int

An Agentic AI Framework Overcomes Fundamental Limitations of Large Language Models for Glaucoma Detection from Fundus Photography

arXiv:2608.07651v1 Announce Type: new Abstract: Large language models (LLMs) show promise in medical image interpretation but suffer from hallucination, limited accuracy, and run-to-run inconsistency.

BASIS: Breach-Aware Selective Prompt Injection Shielding with Prefill Attention Probes

arXiv:2608.08027v1 Announce Type: cross Abstract: Prompt injection is a critical security threat in large language model (LLM) applications, where attackers hijack model behavior by embedding maliciou

researcharxiv-cs-lg

Bounding Hallucinations: Merlin-Arthur Protocols for Mutual-Information Bounds in Language Models

arXiv:2512.11614v3 Announce Type: replace-cross Abstract: Retrieval-augmented generation (RAG) relies on retrieved context to guide large language models (LLM), yet treats the retrieval as a heuristic

ColluSkill: Adversarial Cross-Skill Composition for Evading Agent Skill Scanners

arXiv:2608.09732v1 Announce Type: cross Abstract: Agent skills are emerging as an important attack surface in LLM-based agent systems. Through an empirical study of existing skill scanners, we find th

local-aiarxiv-cs-ai

Coupled Graph--Policy Distillation for Personalized Medication Safety in Older Adults with Multimorbidity

arXiv:2608.09443v1 Announce Type: new Abstract: Large language model (LLM) agents can support medication review between clinical visits, but safe choices for older adults with multimorbidity depend on

How to Ask the AI: A User Perspective Survey for Large Language Model Prompting

arXiv:2608.07494v1 Announce Type: cross Abstract: AI tools like ChatGPT and DeepSeek, powered by Large Language Models (LLMs), allow users to obtain instant and effective content responses simply by t

LatticeMind: A Conflict-Aware Memory Primitive for Multi-Agent Systems

arXiv:2608.08236v1 Announce Type: new Abstract: Multi-agent LLM systems often fail not for lack of candidate answers, but because they have no persistent mechanism for deciding which incompatible clai

Rethinking Factor Sharing in Federated LoRA: A Rank-Aware Adaptive Approach

arXiv:2608.09742v1 Announce Type: cross Abstract: Low-rank adaptation (LoRA) represents large language model (LLM) updates with two compact matrix factors, i.e., A and B, providing an efficient way to

local-aiarxiv-cs-ai

SkillSmith: Enhancing Locally Deployed Agents via Automatic Skill Construction and Evolution

arXiv:2608.08037v1 Announce Type: new Abstract: LLM-based agent frameworks now act as personal assistants for multi-step tasks. Existing agent frameworks such as OpenClaw commonly follow the Cloud Age

local-aiarxiv-cs-ai

take away the symbolic part of this and it just would not have worked. Claude on Riemann is yet another victory for hybrid, neurosymbolic AI…

model-releasesgary-marcus--x

On August 11 2026, Gary Marcus commented that if the symbolic components were removed from Claude’s Riemann implementation it would fail to work, highlighting a concrete win for hybrid neurosymbolic A

Toy project: a chat title model that fits in 5 MiB of ram

model-releasesr-localllama

Not even sure if I'm allowed to post this, what with the 'completely/primarily LLM generated copy' rule (the post itself is fine, but the repo/model I'm sharing definitely is, whoops) and the whole li

Tree-of-Experience: Hierarchical Experience Management for Self-Evolving Agents

arXiv:2608.09044v1 Announce Type: new Abstract: Continual self-evolution requires LLM agents to transform environmental interactions into reliable and reusable experience. Existing methods typically r

Verication-driven closed-loop multi-agent large language modelframework for code-compliant structural design

arXiv:2608.07978v1 Announce Type: cross Abstract: Multi-agent large language model(LLM)systems are applied to structural design,yet most use one-shot generation and cannot verify their output,leaving

Who Verifies the Benchmark? Decentralizing Trust in Large Language Model Evaluation

arXiv:2608.07762v1 Announce Type: new Abstract: LLM benchmarks can build an organization's reputation and attract customers, but only when results are transparent and verifiable. Unverified claims tha

Applications

Autonomy-of-Heads: Data-Free Sparse Attention from Frozen Query-Key Geometry

arXiv:2608.06849v1 Announce Type: cross Abstract: Long-context LLM inference is bottlenecked by quadratic attention computation and growing KV-cache costs. Existing sparse attention and KV-compression

applicationsarxiv-cs-ai

b10338

model-releasesllama-cpp-releases

model-saver : fix expert shared/chunk FFN length key clobber (#26693) The saver called add_kv with LLM_KV_EXPERT_SHARED_FEED_FORWARD_LENGTH twice, the second time passing n_ff_chexp. gguf_set_val_u32

Comparing how Cline, Kilo, and Qwen Code handle long-task context/state (and why context loops keep happening)

model-releasesr-localllama

I've been comparing Cline / Kilo / Qwen Code lately since they all handle long-task state differently. Cline: has Focus Chain, a markdown file kept outside the conversation that gets reinjected on a c

Confirming Our Biases? Evaluating the Capabilities, Risks, and Societal Impact of Large Language Models

arXiv:2608.06977v1 Announce Type: new Abstract: It is well established that large language models (LLMs) are sensitive to prompt framing, reflecting patterns in their training data or prior prompts. I

Applications

Direct Visual Grounding by Directing Attention of Visual Tokens

arXiv:2511.12738v2 Announce Type: replace Abstract: Vision Language Models (VLMs) mix visual tokens and text tokens. A puzzling issue is the fact that visual tokens most related to the query receive l

applicationsarxiv-cs-cv