About
Research & Development
Unlocking Counterfactuals in Systems with Feedback Loops
TLDR: This paper develops a theoretical framework for counterfactual inference in cyclic structural causal models (SCMs) under shift-scale interventions. It proves that SCMs satisfying...
Research & Development
Advancing Psychiatric Diagnosis with AI-Generated Clinical Dialogues for Comorbidity
TLDR: Researchers have developed PsyCoTalk, the first large-scale dataset of 3,000 multi-turn diagnostic dialogues for psychiatric comorbidity. This dataset is created using a novel...
Research & Development
Graph-based Planning: A Leap Forward for AI Agent Efficiency
TLDR: GAP (Graph-based Agent Planning) is a new framework that enables large language model (LLM) agents to perform tasks more efficiently and accurately by...
Research & Development
Dynamic SQL Generation: How MTIR-SQL Enhances Text-to-SQL with Interactive Reasoning
TLDR: MTIR-SQL is a new reinforcement learning framework for Text-to-SQL that uses multi-turn, tool-integrated reasoning. Unlike previous methods, it incorporates real-time database execution feedback...
Research & Development
Expanding AI Reasoning: A New Approach to Zero Reinforcement Learning for Diverse Tasks
TLDR: Researchers from Tencent have introduced a new Zero Reinforcement Learning (Zero-RL) method that significantly improves large language models' (LLMs) ability to reason across...
Research & Development
New Method Enhances Reinforcement Learning Exploration by Generating Critical Experiences
TLDR: A new research paper introduces Modelic Generative Exploration (MoGE), a novel approach to improve off-policy Reinforcement Learning by augmenting exploration. MoGE uses a...