Posts
All the articles I've posted.
- 7.0
Your LLM Agents are Temporally Blind: The Misalignment Between Tool Use Decisions and Human Time Perception
arXiv:2510.23853v3 Announce Type: replace Abstract: Large language model (LLM) agents are increasingly used to interact with and execute tasks in dynamic environments. However, a critical yet overlook
- 6.0
A Mechanistic Account of Attention Sinks in GPT-2: One Circuit, Broader Implications for Mitigation
arXiv:2604.14722v1 Announce Type: new Abstract: Transformers commonly exhibit an attention sink: disproportionately high attention to the first position. We study this behavior in GPT-2-style models w
- 5.5
Cosine-Similarity Routing with Semantic Anchors for Interpretable Mixture-of-Experts Language Models
arXiv:2509.14255v2 Announce Type: replace Abstract: Mixture-of-Experts (MoE) models improve efficiency through sparse activation, but their learned gating functions provide limited insight into routin
- 6.5
The Mirror Design Pattern: Strict Data Geometry over Model Scale for Prompt Injection Detection
arXiv:2603.11875v2 Announce Type: replace-cross Abstract: Prompt injection defenses are often framed as semantic understanding problems and delegated to increasingly large neural detectors. For the fi
- 5.5
Theory of Mind in Action: The Instruction Inference Task in Dynamic Human-Agent Collaboration
arXiv:2507.02935v2 Announce Type: replace Abstract: Successful human-agent teaming relies on an agent being able to understand instructions given by a (human) principal. In many cases, an instruction
- 5.5
In-Context Autonomous Network Incident Response: An End-to-End Large Language Model Agent Approach
arXiv:2602.13156v2 Announce Type: replace-cross Abstract: Rapidly evolving cyberattacks demand incident response systems that can autonomously learn and adapt to changing threats. Prior work has exten
- 6.0
Enhancing LLM-based Search Agents via Contribution Weighted Group Relative Policy Optimization
arXiv:2604.14267v1 Announce Type: new Abstract: Search agents extend Large Language Models (LLMs) beyond static parametric knowledge by enabling access to up-to-date and long-tail information unavaila
- 6.0
GUI-Perturbed: Domain Randomization Reveals Systematic Brittleness in GUI Grounding Models
arXiv:2604.14262v1 Announce Type: new Abstract: GUI grounding models report over 85% accuracy on standard benchmarks, yet drop 27-56 percentage points when instructions require spatial reasoning rathe
- 6.5
Awakening Dormant Experts:Counterfactual Routing to Mitigate MoE Hallucinations
arXiv:2604.14246v1 Announce Type: new Abstract: Sparse Mixture-of-Experts (MoE) models have achieved remarkable scalability, yet they remain vulnerable to hallucinations, particularly when processing
- 5.5
MixAtlas: Uncertainty-aware Data Mixture Optimization for Multimodal LLM Midtraining
arXiv:2604.14198v1 Announce Type: new Abstract: Domain reweighting can improve sample efficiency and downstream generalization, but data-mixture optimization for multimodal midtraining remains largely
- 7.0
CoopEval: Benchmarking Cooperation-Sustaining Mechanisms and LLM Agents in Social Dilemmas
arXiv:2604.15267v1 Announce Type: cross Abstract: It is increasingly important that LLM agents interact effectively and safely with other goal-pursuing agents, yet, recent works report the opposite tr
- 5.5
OpenMobile: Building Open Mobile Agents with Task and Trajectory Synthesis
arXiv:2604.15093v1 Announce Type: cross Abstract: Mobile agents powered by vision-language models have demonstrated impressive capabilities in automating mobile tasks, with recent leading models achie
- 5.5
RaTA-Tool: Retrieval-based Tool Selection with Multimodal Large Language Models
arXiv:2604.14951v1 Announce Type: cross Abstract: Tool learning with foundation models aims to endow AI systems with the ability to invoke external resources -- such as APIs, computational utilities,
- 7.0
Between a Rock and a Hard Place: The Tension Between Ethical Reasoning and Safety Training in LLMs
arXiv:2509.05367v4 Announce Type: replace-cross Abstract: Large Language Model safety alignment predominantly operates on a binary assumption that requests are either safe or unsafe. This classificati
- 6.5
The LLM Fallacy: Misattribution in AI-Assisted Cognitive Workflows
arXiv:2604.14807v1 Announce Type: cross Abstract: The rapid integration of large language models (LLMs) into everyday workflows has transformed how individuals perform cognitive tasks such as writing,
- 6.5
Prompt Optimization Is a Coin Flip: Diagnosing When It Helps in Compound AI Systems
arXiv:2604.14585v1 Announce Type: cross Abstract: Prompt optimization in compound AI systems is statistically indistinguishable from a coin flip: across 72 optimization runs on Claude Haiku (6 methods
- 6.0
Don't Retrieve, Navigate: Distilling Enterprise Knowledge into Navigable Agent Skills
arXiv:2604.14572v1 Announce Type: cross Abstract: Retrieval-Augmented Generation (RAG) grounds LLM responses in external evidence but treats the model as a passive consumer of search results: it never
- 6.0
MARS²: Scaling Multi-Agent Tree Search via Reinforcement Learning for Code Generation
arXiv:2604.14564v1 Announce Type: cross Abstract: Reinforcement learning (RL) paradigms have demonstrated strong performance on reasoning-intensive tasks such as code generation. However, limited traj
- 6.0
Dissecting Failure Dynamics in Large Language Model Reasoning
arXiv:2604.14528v1 Announce Type: cross Abstract: Large Language Models (LLMs) achieve strong performance through extended inference-time deliberation, yet how their reasoning failures arise remains p
- 5.5
From Tokens to Steps: Verification-Aware Speculative Decoding for Efficient Multi-Step Inference
arXiv:2604.15244v1 Announce Type: new Abstract: Speculative decoding (SD) accelerates large language model inference by allowing a lightweight draft model to propose outputs that a stronger target mod