Knowledge catalogue
arXiv:2605.22001v1 Announce Type: cross Abstract: Injection detectors deployed to protect LLM agents are calibrated on static, template-based payloads that announce themselves as override directives.
arXiv:2605.20923v1 Announce Type: cross Abstract: Distributed LLM agent workflows should not be monitored as if they produced a single sequential log. In an asynchronous execution, a decision can only
arXiv:2602.08023v3 Announce Type: replace-cross Abstract: Existing benchmarks for LLM-based offensive security agents use isolated, single-target setups with a known vulnerable service and fixed objec
arXiv:2605.21958v1 Announce Type: new Abstract: When a multi-module LLM agent fails, the module most responsible for the failure is not necessarily the best place to intervene. We demonstrate this Dia
arXiv:2605.22074v1 Announce Type: cross Abstract: Reinforcement learning from verifiable rewards (RLVR) has shown strong promise for LLM reasoning, but outcome-based RLVR remains inefficient on hard p
arXiv:2605.21405v1 Announce Type: cross Abstract: Third-party Python libraries introduce dependency management overhead, supply chain risk, and deployment friction in constrained environments. A natur
arXiv:2605.21240v1 Announce Type: new Abstract: LLM agents have shown strong performance across a wide range of complex tasks, including interactive environments that require long-horizon decision mak
If this is true, using the best public estimates we have of LLM resource use, solving this Erdos problem took 0.6–6.3 kWh of electricity and about 3–31 liters of water. So that is less than three almo
arXiv:2605.21347v1 Announce Type: cross Abstract: Diagnosing failures in LLM agents remains largely manual. Practitioners inspect a small subset of execution traces, form ad-hoc hypotheses, and iterat
arXiv:2605.20285v1 Announce Type: new Abstract: We tackle the question of how to scale more efficiently across the many, ever-growing stages of current LLM training pipelines. Our guiding intuition st
LLM training is built on fast MatMuls. But many surrounding ops still run as memory-bound kernels. CODA reparameterizes them to hide in the matmul’s shadow, fused into its epilogue before results leav
arXiv:2605.20833v1 Announce Type: new Abstract: Memory is a central capability for LLM agents operating across long-horizon tasks. Existing memory benchmarks predominantly evaluate retention of person
arXiv:2605.21404v1 Announce Type: new Abstract: We read twelve well-known LLM agent benchmark papers and recorded, dimension by dimension, what each paper actually says about how its evaluation was ru
arXiv:2605.18815v1 Announce Type: new Abstract: Modern large language model (LLM) training is inherently dynamic: resource fluctuations, RLHF phase shifts, and cluster elasticity continually reshape t
arXiv:2605.19604v1 Announce Type: new Abstract: Large Language Model (LLM) agents increasingly act inside real workspaces, where tools and skills determine whether model reasoning becomes reliable act
arXiv:2603.11768v2 Announce Type: replace Abstract: Long-term memory has emerged as a foundational component of autonomous Large Language Model (LLM) agents, enabling continuous adaptation, lifelong m
arXiv:2602.04998v2 Announce Type: replace-cross Abstract: Low-Rank Adaptation (LoRA) is the prevailing approach for efficient large language model (LLM) fine-tuning. Building on this paradigm, recent
LLM agents & memory systems operate in continuously updated environments (Git repos, evolving docs). They must process long contexts, recover earlier information, and reason over many updates that cre
// Memory as a Model // The paper augments any LLM with a separate trained memory model that stores, retrieves, and integrates facts on its behalf. It decouples memory updates from base-model weight u
arXiv:2605.19127v1 Announce Type: new Abstract: LLM agents increasingly have access to private user data and act on the user's behalf when interacting with third-party systems. The user defines what m
arXiv:2605.18805v1 Announce Type: cross Abstract: LLM recommendation agents increasingly produce structured recommendation reports: sets of items accompanied by natural-language justifications. Yet ex
arXiv:2605.18882v1 Announce Type: cross Abstract: LLM agents exhibit a consistent tendency to over-call, invoking tools even in situations where none is needed. On the When2Call benchmark, six models
arXiv:2605.18796v1 Announce Type: cross Abstract: LLM cascades and model routing promise lower inference cost by sending easy queries to a small model and escalating hard ones to a large model, but mo
arXiv:2602.03955v2 Announce Type: replace Abstract: While large language model (LLM) multi-agent systems achieve superior reasoning performance through iterative debate, practical deployment is limite
arXiv:2605.17324v1 Announce Type: cross Abstract: Clarification-seeking behavior is widely regarded as a desirable property of LLM agents, enabling them to resolve ambiguity before acting on underspec
arXiv:2605.17641v1 Announce Type: new Abstract: Long-horizon LLM agents rely on persistent memory to support interactions across sessions, yet existing memory systems often retrieve context using sema
arXiv:2605.17304v1 Announce Type: cross Abstract: LLM context is not just tokens; it is a set of commitments. Long-running conversations accumulate goals, constraints, decisions, preferences, tool res
arXiv:2605.17281v1 Announce Type: cross Abstract: Tool-augmented LLM agents call APIs whose intermediate outputs, such as presigned URLs, session tokens, and OAuth state parameters, are observation co
arXiv:2605.16826v1 Announce Type: cross Abstract: Knowledge distillation is central to LLM post-training, yet its design space remains poorly understood, especially alongside reinforcement learning (R
arXiv:2605.16336v1 Announce Type: cross Abstract: Large language models (LLMs) have made fluent essay writing, code drafting, and quiz answering instantly available to students at every level, from se
arXiv:2601.07122v2 Announce Type: replace-cross Abstract: While virtualization and resource pooling empower cloud networks with structural flexibility and elastic scalability, they inevitably expand t
arXiv:2602.16346v3 Announce Type: replace Abstract: LLM-based agents execute real-world workflows via tools and memory. These affordances enable ill-intended adversaries to also use these agents to ca
arXiv:2503.02161v3 Announce Type: replace Abstract: Synthetic tabular data are increasingly being used to replace real data, serving as an effective solution that simultaneously protects privacy and a
arXiv:2605.17596v1 Announce Type: new Abstract: We present NeuSymMS, an adaptive memory system that enables large language model (LLM) agents to learn, remember, and reason about users across sessions
arXiv:2605.16858v1 Announce Type: cross Abstract: Autonomous Vehicles (AVs) must make reliable decisions in dense urban environments where pedestrian behavior is variable, sometimes abnormal, and ofte
arXiv:2605.18199v1 Announce Type: cross Abstract: The rapid growth of tabular datasets in data lakes, data spaces, and open data portals makes effective dataset search essential for reuse and analysis
arXiv:2605.18474v1 Announce Type: cross Abstract: The widespread deployment and redistribution of large language models (LLMs) have made model provenance tracking a critical challenge. While existing
arXiv:2605.17830v1 Announce Type: new Abstract: Safety evaluations of memory-equipped LLM agents typically measure within-task safety: whether an agent completes a single scenario safely, often under
arXiv:2605.17076v1 Announce Type: cross Abstract: Concurrent LLM agents sharing mutable natural-language state produce Structural Race Conditions (SRCs): write-write and cross-shard stale-read conflic
arXiv:2605.18693v1 Announce Type: new Abstract: As LLM agents are increasingly built around reusable skills, a central challenge is no longer only whether agents can use provided skills, but whether t
arXiv:2605.16986v1 Announce Type: cross Abstract: LLM agents benefit from reusable skills, yet test-time tasks often require guidance more specific than a static skill library can provide. We propose
arXiv:2605.16746v1 Announce Type: new Abstract: LLM agents increasingly rely on persistent state, including transcripts, summaries, retrieved context, and memory buffers, to support long-horizon inter
Today we release Contrastive Neuron Attribution (CNA), a method for steering LLM behavior by identifying and ablating sparse circuits in the MLP basis without training a sparse autoencoder, modifying
arXiv:2605.17453v1 Announce Type: cross Abstract: Tool-using LLM agents increasingly rely on external tools to make consequential decisions, yet most existing agent-security benchmarks and defenses im
This post explores conceptual parallels between classical distributed systems design patterns—such as actor models and reactive blackboard architectures—and their potential application to LLM agent de
arXiv:2605.15889v1 Announce Type: cross Abstract: Security in cloud computing has become a major concern due to several factors such as layered cloud architectures, dynamic environments, and exposure
arXiv:2605.15343v1 Announce Type: new Abstract: LLM-based agents are increasingly used to simulate deliberative interactions such as negotiation, conflict resolution, and multi-turn opinion exchange.
arXiv:2512.00417v5 Announce Type: replace Abstract: This paper introduces CryptoBench, the first expert-curated, dynamic benchmark designed to rigorously evaluate the real-world capabilities of Large
arXiv:2605.16194v1 Announce Type: cross Abstract: LLM agents routinely serve as first (and sometimes only) readers of academic papers, skimming for sub-claims, extracting reproducibility steps, and ge
arXiv:2512.08964v4 Announce Type: replace Abstract: Graph learning aims to convert data into graph representations, which are fundamental to many problems in machine learning for CAD, where circuits,
arXiv:2605.15207v1 Announce Type: new Abstract: Multi-agent LLM systems have shown promise for complex reasoning, yet recent evaluations reveal they often underperform single-model baselines. We ident
1/3 My time at Meta FAIR will soon come to a close. I joined nearly two years ago full-time to help advance LLM reasoning. It has been a remarkable journey working with and leading an exceptionally ta
Gary Marcus expresses optimism about the potential of LLM-powered AI agents in a post on X (formerly Twitter). The post advocates for confidence in these AI systems, though without access to the full
New article: a visual tour of recent LLM architecture advances, from Gemma 4 to DeepSeek V4. I focus on long-context efficiency tweaks like KV sharing, per-layer embeddings, layer-wise attention budge
A breakdown of a ComfyUI workflow that combines Seedance 2.0 with an LLM prompt setup designed for cinematic motion shots like this. This example recreates the viral floating hot sauce effect with a M
arXiv:2605.14892v1 Announce Type: new Abstract: LLM-based autonomous agents have demonstrated strong capabilities in reasoning, planning, and tool use, yet remain limited when tasks require sustained
arXiv:2603.24586v2 Announce Type: replace-cross Abstract: As LLMs are increasingly used as judges in code applications, they should be evaluated in realistic interactive settings that capture partial
arXiv:2605.14801v1 Announce Type: new Abstract: Zero-shot vision-and-language navigation (VLN) has gained significant attention due to its minimal data collection costs and inherent generalization. Th
arXiv:2605.14034v1 Announce Type: new Abstract: Wide applications of LLM-based agents require strong alignment with human social values. However, current works still exhibit deficiencies in self-cogni
arXiv:2605.14498v1 Announce Type: new Abstract: Large Language Model (LLM) agents increasingly serve as personal assistants and workplace collaborators, where their utility depends on memory systems t