The Morning
From the arXiv
$A^2E$ : An End-to-End Agent Auditing Engine
A^2E$ is an end-to-end engine for evaluating LLM agents deployed via harnesses. It uses the Agent Task Protocol (ATP) for easy integration and an instrumented Monitor to generate standardized execution traces. This allows for systematic, multidimensional capability assessment beyond simple correctness, offering a more comprehensive evaluation of agent performance.
Adapting Technical-Service LLM Agents with Latent Logic Augmentation, Robust Noise Reduction, and Hybrid Reward Modeling
This paper introduces a novel framework for adapting technical-service LLM agents to production environments. It addresses challenges like hidden decision logic and noisy data by augmenting training with latent logic, robustly filtering noise, and employing a …
Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models
This paper demonstrates that finetuning LLMs, even on unrelated data, can bypass safety mechanisms and enable verbatim recall of copyrighted books. By training models to expand plot summaries, the researchers successfully induced models like GPT-4o and Gemini-…
Alpha-R1: Alpha Screening with LLM Reasoning via Reinforcement Learning
Alpha-R1 uses a reinforcement learning-aligned LLM to screen investment factors. Its core method, "semantic gating," dynamically assesses factors' economic rationale against current market conditions, selecting a sparse, context-aware subset. This approach add…
AMA: Adaptive Memory via Multi-Agent Collaboration
AMA introduces a novel framework for LLM agent memory that addresses limitations of existing systems. Its core method involves a multi-agent collaboration to manage memory across multiple granularities, dynamically aligning retrieval granularity with task comp…
Breaking Planner Integrity Boundary: Enviroment State-Text Injection Attack on LLM-Driven Embodied Agents
This paper introduces the Environment State-Text Injection (ESTI) attack, a novel method that targets LLM-driven embodied agents by manipulating their understanding of the environm…
Clarify, Abstain or Answer? Strategising in Conversation with Belief-Augmented Generation
This paper introduces Belief-Augmented Generation (BAG), a method that leverages an LLM's internal uncertainty (represented by multiple generated responses) to inform its conversat…
Compressing Sequences in the Latent Embedding Space: $K$-Token Merging for Large Language Models
This paper introduces K-Token Merging, a novel method for compressing long prompts in Large Language Models by operating in the latent embedding space. It merges contiguous blocks …
DSAEval: Evaluating Data Science Agents on a Wide Range of Real-World Data Science Problems
DSAEval introduces a comprehensive benchmark for evaluating data science agents on real-world problems. Its core method involves a multimodal environment with multi-query interacti…
FinAcumen: Financial Multimodal Reasoning via Self-Evolving Experience Memory Harness
FinAcumen addresses the limitations of stateless financial reasoning agents by introducing a novel self-evolving experience memory. This memory bank stores and selectively retrieve…
The Town Square
Fable 5.1, an AI model, has successfully deciphered the Cyphral Distich, a complex 370-year-old cipher, marking a significant breakthrough in cryptanalysis.
Workshops
Ever Gauzy is an open-source business management platform offering integrated ERP, CRM, HRM, ATS, and PM functionalities to streamline operations.
This repository provides a real-time, open-source spatial intelligence platform that functions as a spy satellite simulator in your browser, displaying live data on a photorealistic 3D globe.
