AI Agents

AWS turns its S3 storage service into a file system for AI agents

Moody’s Integrates AI Agents With Anthropic’s Claude

Started a video series on building an orchestration layer for LLM post-training [P]

Hi everyone! Context, motivation, a lot of yapping, feel free to skip to TL;DR. A while back I posted here asking [D] What framework do y...

[2602.15654] Zombie Agents: Persistent Control of Self-Evolving LLM Agents via Self-Reinforcing Injections

This paper discusses the security vulnerabilities of self-evolving LLM agents, introducing the concept of 'Zombie Agents' that can be cov...

[2504.05654] Curved representational Bregman divergences and their applications

This article introduces curved representational Bregman divergences, exploring their mathematical foundations and applications in informa...

[2602.15600] The geometry of online conversations and the causal antecedents of conflictual discourse

This article explores the dynamics of conflictual discourse in online conversations, particularly focusing on climate change discussions....

[2602.15564] Beyond Static Pipelines: Learning Dynamic Workflows for Text-to-SQL

The paper presents a novel approach to Text-to-SQL systems by introducing dynamic workflows that adapt during inference, enhancing perfor...

[2602.11618] How Well Do Large-Scale Chemical Language Models Transfer to Downstream Tasks?

This paper evaluates the effectiveness of large-scale Chemical Language Models (CLMs) in transferring knowledge to downstream molecular p...

[2602.10706] Reducing Estimation Uncertainty Using Normalizing Flows and Stratification

This paper presents a novel approach to reducing estimation uncertainty in statistical analysis using normalizing flows and stratified sa...

[2602.08032] Horizon Imagination: Efficient On-Policy Rollout in Diffusion World Models

The paper presents Horizon Imagination (HI), an innovative on-policy imagination process for reinforcement learning using diffusion-based...

[2602.07418] Achieving Optimal Static and Dynamic Regret Simultaneously in Bandits with Deterministic Losses

This paper presents an algorithm that achieves optimal static and dynamic regret simultaneously in adversarial multi-armed bandits with d...

[2602.15513] Improving MLLMs in Embodied Exploration and Question Answering with Human-Inspired Memory Modeling

This paper presents a novel non-parametric memory framework for improving Multimodal Large Language Models (MLLMs) in embodied exploratio...

[2602.05999] On the Role of Iterative Computation in Reinforcement Learning

This paper explores how the amount of compute available to reinforcement learning (RL) policies influences their learning capabilities an...

[2601.19720] Improving Policy Exploitation in Online Reinforcement Learning with Instant Retrospect Action

The paper presents a novel algorithm, Instant Retrospect Action (IRA), aimed at enhancing policy exploitation in online reinforcement lea...

[2601.10498] PROMA: Projected Microbatch Accumulation for Reference-Free Proximal Policy Updates

The paper introduces Projected Microbatch Accumulation (PROMA), a novel method for proximal policy updates that enhances KL divergence co...

[2602.15439] Algorithmic Approaches to Opinion Selection for Online Deliberation: A Comparative Study

This article examines various algorithmic approaches to opinion selection in online deliberation, highlighting the trade-offs between div...

[2602.15397] ActionCodec: What Makes for Good Action Tokenizers

The paper introduces ActionCodec, a novel action tokenizer designed to enhance Vision-Language-Action (VLA) models by optimizing tokeniza...

[2602.15377] Orchestration-Free Customer Service Automation: A Privacy-Preserving and Flowchart-Guided Framework

This paper presents an orchestration-free framework for customer service automation, utilizing Task-Oriented Flowcharts (TOFs) to enhance...

[2602.15362] Automated Multi-Source Debugging and Natural Language Error Explanation for Dashboard Applications

This paper presents a novel system for Automated Multi-Source Debugging and Natural Language Error Explanation, aimed at improving user e...

[2602.15353] NeuroSymActive: Differentiable Neural-Symbolic Reasoning with Active Exploration for Knowledge Graph Question Answering

The paper presents NeuroSymActive, a novel framework for Knowledge Graph Question Answering that integrates differentiable neural-symboli...

[2602.15350] Fine-Tuning LLMs to Generate Economical and Reliable Actions for the Power Grid

This paper discusses a method for fine-tuning large language models (LLMs) to generate effective corrective actions for power grid manage...

[2510.26792] Learning Pseudorandom Numbers with Transformers: Permuted Congruential Generators, Curricula, and Interpretability

This article explores how Transformer models can learn sequences generated by Permuted Congruential Generators (PCGs), demonstrating thei...

[2510.03269] General Exploratory Bonus for Optimistic Exploration in RLHF

This paper introduces the General Exploratory Bonus (GEB) framework, which enhances optimistic exploration in reinforcement learning with...