The Morning
From the arXiv
Break It Down, Pass It On: Cross-Task Skill Transfer in LLM Agents
his paper investigates how LLM agents learn and reuse skills across tasks. The core method involves comparing different skill induction strategies: task-level vs. subtask-level and text vs. code formats. The key contribution is demonstrating that subtask-level skill induction and text-based skill representation lead to more reliable and beneficial skill transfer, improving agent performance compared to task-level induction or code-based skills.
Optimal Skill Selection for LLM Agents with Provable Bicriteria Guarantees
This paper addresses the challenge of selecting optimal skills for LLM agents within a limited context window. The authors propose a novel optimization framework that balances skill benefit against token cost, moving beyond independent skill scoring. Their dev…
FormalTCS: Benchmarking End-to-End Frontier Formal Theoretical Computer Science Research of Large Language Models
This paper introduces FormalTCS, an expert-validated benchmark for evaluating LLMs on realistic, end-to-end theoretical computer science research tasks. It comprises 175 instances from top TCS conferences, preserving original definitions and proof structures w…
AI4AI-Bench: Benchmarking LLM Agents in Algorithmic Design for Recursive Self-Improvement
This paper introduces AI4AI-Bench, a novel benchmark designed to evaluate Large Language Model (LLM) agents' ability to design training algorithms for recursive self-improvement (RSI). The core method involves agents modifying existing training algorithms with…
DARS: Dual-Level Credit Assignment RL with Structured Reasoning for Instruction-Based Image Editing
This paper introduces DARS, a reinforcement learning framework for instruction-based image editing. DARS addresses the inefficiency of traditional training by employing dual-level credit assignment. It achieves this by using multi-plan, multi-render rollouts t…
From Agent Behaviour to Agent-Friendly Documentation: An Empirical Study of How Coding Agents Discover, Read, and Write Technical Documentation
This paper empirically studies how coding agents interact with technical documentation. It reveals that agents primarily consult agent-facing artifacts like instruction files and w…
Inject, Align, Recover: Staged Post-Training for Retrieval-Free Document Knowledge Internalization
This paper introduces IAR, a three-stage post-training method for enabling large language models to answer questions from a fixed document set without retrieval. IAR first injects …
Let's Scale Step by Step: Compute-Efficient Hyperparameter Transfer for Large-Scale Mixture-of-Experts
This paper introduces a compute-efficient method to transfer optimal learning rates for large Mixture-of-Experts (MoE) models. It first demonstrates that optimal learning rates are…
Manifold Drift in Flow Preference Optimization: A Root Cause of Reward Hacking
This paper identifies "manifold drift" as a key cause of reward hacking in continuous-time generative model alignment. The authors show that standard preference optimization can pu…
MemTrapBench: Benchmarking Cognitive Traps in LLM Memory Use
This paper introduces MemTrapBench, a novel benchmark designed to evaluate how retrieved memories can negatively impact LLM reasoning and performance, a phenomenon termed "cognitiv…
The Town Square
AI companies are destroying physical books, prompting a call to scan rare books before they are lost forever.
Workshops
This repository provides a curated collection of "skills" or reusable code snippets, inspired by the author's personal `.agents` directory, designed for practical application by real engineers.
OpenLogi is a native, local-first Rust application that provides an account-free, telemetry-free alternative to Logitech Options+, enabling button remapping, DPI adjustments, and SmartShift control via HID++.
