The Morning
From the arXiv
AgentOrchestra: Orchestrating Multi-Agent Intelligence with the Tool-Environment-Agent(TEA) Protocol
his paper introduces the Tool-Environment-Agent (TEA) protocol, a novel framework for coordinating multi-agent systems. TEA models agents, tools, and environments as versioned resources with explicit lifecycles, enabling better context management and reproducibility. Building on TEA, AgentOrchestra provides a hierarchical framework for dynamic agent capability extension and coordination on complex tasks.
BenchTrace: A Benchmark for Testing Reflection Ability and Controlled Evolution in LLM Agents
This paper introduces BenchTrace, a benchmark designed to evaluate the reflection and controlled evolution capabilities of LLM agents. BenchTrace utilizes a dataset of annotated episodes and includes two key evaluations: one that probes failure identification …
EAPO: Enhancing Policy Optimization with On-Demand Expert Assistance
EAPO is a novel RL framework that enhances LLM reasoning by allowing policies to adaptively seek assistance from external experts during training. This on-demand expert interaction provides richer reward signals and more reliable reasoning trajectories, ultima…
Estimating the Empowerment of Language Model Agents
This paper introduces EELMA, an algorithm that estimates the "empowerment" of language model agents. Empowerment, an information-theoretic measure, quantifies an agent's ability to influence future states through its actions. EELMA allows for scalable evaluati…
EvoSpec: Evolving Speculative Decoding via Real-Time Vocabulary and Parameter Adaptation
EvoSpec accelerates LLM inference by dynamically adapting the draft model's vocabulary and parameters in real-time. This approach overcomes the limitations of static pruning by efficiently retrieving relevant long-tail tokens and minimizing the distributional …
From AR to Diffusion: Efficiently Adapting Large Language Models with Strictly Causal and Elastic Horizons
This paper introduces FLUID, a framework for efficiently adapting autoregressive (AR) language models to diffusion models for text generation. FLUID achieves this by enforcing "Str…
From Meta-Thought to Execution: Cognitively Aligned Post-Training for Generalizable and Reliable LLM Reasoning
This paper proposes a new LLM post-training method called Chain-of-Meta-Thought (CoMT). It addresses the limitation of current methods by mimicking human problem-solving, which inv…
Graph-Enhanced Policy Optimization in LLM Agent Training
This paper introduces Graph-Enhanced Policy Optimization (GEPO) to improve LLM agent training for multi-step tasks. GEPO addresses the issue of uniform credit assignment by develop…
GroundAct: Can LLM Agents Ground Actions in Environmental States?
This paper introduces GroundAct, a benchmark designed to assess Large Language Model (LLM) agents' ability to ground actions in environmental states. The core method involves evalu…
InsightEval: An Expert-Curated Benchmark for Assessing Insight Discovery in LLM-Driven Data Agents
This paper addresses the lack of robust benchmarks for evaluating LLM-driven data agents' insight discovery capabilities. The authors identify critical flaws in existing benchmarks…
The Town Square
Google has released Gemma 4 12B, a unified, encoder-free multimodal model designed for developers, capable of processing and generating both text and images.
Workshops
This repository provides an agent harness for performance optimization, focusing on skills, instincts, memory, security, and research-first development for various AI coding tools like Claude Code and Codex.
This Claude Code skill acts like a caveman, using fewer words to achieve the same goal, thereby reducing token usage by up to 65%.
