The Morning
From the arXiv
EDGE: Error Dependency Graph-Guided Multi-Error Attribution in Multi-Agent LLM Systems
his paper introduces EDGE, a framework that addresses multi-agent LLM failures by explicitly modeling dependencies between multiple errors. EDGE constructs an error dependency graph and uses counterfactual reasoning to identify a reliable causal subset of errors. This graph-guided approach improves the accuracy of multi-error attribution, leading to better explanations and repair strategies.
EmbodiedSkills: A Unified Framework for Orchestrating, Training, and Deploying VLA Agents
EmbodiedSkills is a unified framework for Vision-Language-Action (VLA) agents that treats skill decisions as executable proposals. It introduces a runtime that verifies skill prerequisites before execution and checks outcomes afterward, ensuring robust long-ho…
Evaluating Multimodal LLMs as Generalist Vision-Language-Action Agents for Drone Control: Commanding, Approaching, Tracking and Searching
This paper evaluates Multimodal Large Language Models (MLLMs) as generalist agents for drone control by directly integrating them into the control loop with a prompt-defined action space. Their contribution is DroneCATS-Agent, a flexible architecture, and Dron…
GlossoGen: Emergent Language in Complex Multi-Agent LLM Interactions
GlossoGen is a novel platform designed to study language evolution in complex multi-agent LLM interactions. The paper demonstrates that LLM agents, under pressure and with partial information, can develop compositional and morphologically productive languages …
Harness-of-Harness: Multi-Day Autonomous Software Development with Continual Improvement
This paper introduces Harness-of-Harness (HoH), a framework for multi-day autonomous software development using LLM coding agents. HoH enables continual improvement by organizing agents into iterative planning-coding-testing loops, balancing repair with capabi…
The Rise of Verbal Reinforcement Learning
This paper introduces Verbal Reinforcement Learning (VRL), a paradigm where natural language serves as the feedback mechanism for language agents. It unifies existing research by c…
When Safety Routing Breaks: Understanding Alignment Fragility under Benign Fine-Tuning
This paper argues that safety alignment in LLMs is fragile due to a "routing pathway" in output-side MLP modules that is selectively re-sharpened by benign fine-tuning. This Fisher…
HarnessDev: Can LLMs Create and Evolve Their Own Agent Harness?
This paper introduces HarnessDev, a benchmark that evaluates an LLM's ability to create and improve its own agent execution infrastructure (harness). Instead of just assessing task…
A Composable Evaluation System for Reproducible Omni-Modal Foundation Model Evaluation
This paper introduces OmniEvaluator, a system that addresses the challenge of evaluating omni-modal foundation models by unifying existing, incompatible evaluation toolkits. Its co…
CordisBench: Can Language Models Reason About Component Lifecycles in Dynamic Agent Harnesses?
This paper introduces CordisBench, a benchmark designed to test language models' ability to reason about component lifecycles in dynamic agent systems where models can modify their…
The Town Square
Ed Zitron's AI skeptic predictions have been largely accurate, highlighting the hype and potential downsides of AI development more effectively than many optimistic forecasts.
Workshops
This repository provides a structured framework for academic research, guiding users through the iterative process of research, writing, review, revision, and finalization.
OpenMAIC is an immersive, multi-agent learning platform that provides a one-click solution for interactive educational experiences.
