The Morning
From the arXiv
Adaptive Adversaries: A Multi-Turn, Multi-LLM Benchmark for LLM Agent Security
his paper introduces a novel benchmark for evaluating LLM agent security against adaptive, multi-turn attacks. Its core method involves an autonomous LLM attacker that learns from defender responses across multiple rounds to refine its attacks. The key contribution is demonstrating that adaptive, multi-LLM attacks significantly increase success rates compared to static attacks, uncovering novel vulnerabilities and highlighting the limitations of current defense strategies.
Exploratory and Assimilating Reflection: Reflective Recall Cycle for Long-term Memory
This paper introduces Exploratory-Assimilating Reflection (EAR), a novel framework for LLM-based agents to manage long-term memory. EAR employs an iterative search (Exploratory Reflection) to gather relevant memories and then refines a global reranker by repla…
How Does Alignment Tuning Shape Representations of Sycophancy and Related Cue-Induced Biases in LLMs?
This paper investigates how alignment tuning in LLMs creates susceptibility to prompt-induced biases like sycophancy. The core method involves extracting and analyzing "bias directions" within model representations. The key contribution is demonstrating that a…
LLMs and Agentic AI Systems for Smart Grids: A Tutorial on Architectures and Applications
This paper proposes a "solver-grounded" design principle for LLM-based agentic AI systems in smart grids, ensuring numerical outputs are from trusted tools and explicitly verified. This approach addresses the lack of unified design and evaluation in the field,…
MADA-RL: Multi-Agent Debate-Aware Reinforcement Learning for Parameter-Efficient Reasoning in Compact Models
MADA-RL is a post-training framework that trains compact language models for efficient reasoning. It specializes models into generator and critic roles, using a novel "counterfactual critic advantage" signal to train the critic to improve upon the generator's …
Rethinking Heterogeneous LLM Merging: A Weighted Model Averaging Perspective
This paper proposes a simple, training-free method for merging heterogeneous Large Language Models (LLMs) by adapting their parameter spaces and then performing weighted averaging.…
DiFA: Inference-Time Forward-Process Alignment for Diffusion Models
DiFA reframes diffusion model inference as a sequential state estimation problem, treating iterative predictions as correlated observations. It builds a "forward-aligned temporal c…
PPL-Factory: Task-Aware and Budget-Aware Data Selection from Language Modeling to Reasoning
PPL-Factory proposes a task-aware and budget-aware data selection method for fine-tuning large language models. It improves upon existing perplexity-based approaches by considering…
A Geometric Perspective on Stabilizing Value Conflict Resolution
This paper proposes using Chain-of-Thought (CoT) reasoning to stabilize value conflict resolution in LLMs, which are often hindered by compressed scalar rewards in RLHF. Geometrica…
Can We Break LLMs Out of Self-Loops? Fine-Grained Reasoning Control with Activation Steering
This paper introduces SOPHIA, a method to control the reasoning process of LLMs by intervening on their internal latent states. It identifies that LLMs can get stuck in "self-loops…
The Town Square
China's open-weights AI approach is gaining traction, contrasting with the US's proprietary model, suggesting a strategic advantage for China in AI development.
Workshops
This repository builds a local-first code intelligence graph to optimize AI coding tools by creating a persistent, benchmarked map of your codebase, reducing context for reviews and large-repo workflows.
This repository provides an intelligent agent harness for code, enabling sophisticated code generation, analysis, and manipulation through advanced AI capabilities.
