About
Research & Development
Decoding AI’s Thought Process: The Role of Internal States in Reasoning Models
TLDR: This research investigates the internal mechanisms of reasoning language models, specifically focusing on the 'wait' token, which often signals complex reasoning behaviors like...
Research & Development
How AI Agents Learn and Fix Their Internal Maps
TLDR: This research introduces LLM-MapRepair, a framework that enables AI language models to build and maintain accurate spatial maps from text descriptions. It addresses...
Research & Development
AGENTRL: A Scalable Framework for Training Generalist LLM Agents
TLDR: AGENTRL is a new framework that enables large language models (LLMs) to learn more effectively in complex, multi-turn, and multi-task environments. It features...
Research & Development
Faster AI Agents: A Framework for Parallel Execution
TLDR: AI agents are often slow due to sequential API calls. The "Speculative Actions" framework addresses this by using a fast "Speculator" model to...
Research & Development
Unmasking AI Agent Fragility: A New Method for Realistic User Behavior Simulation
TLDR: AI agents struggle with varied human behaviors like impatience or confusion, despite performing well on standard tests. Researchers introduce TraitBasis, a method to...
Research & Development
Code-Based AI Models Master Diverse Games
TLDR: Researchers from Google DeepMind developed Code World Models (CWMs), an approach where Large Language Models (LLMs) translate natural language game rules and gameplay...