TheWorthyOne/

OpenManus-RL

	Dataset Card for OpenManusRL




	
	Dataset Description




	
	Overview

💻 [Github Repo]

OpenManusRL combines agent trajectories from AgentInstruct, Agent-FLAN and AgentTraj-L(AgentGym) with features:

🔍 ReAct Framework - Reasoning-Acting integration 🧠 Structured Training - Separate format/reasoning learning 🚫 Anti-Hallucination - Negative samples + environment grounding 🌐 6 Domains - OS, DB, Web, KG, Household, E-commerce

	Dataset Overview




	

Source… See the full description on the dataset page: https://huggingface.co/datasets/TheWorthyOne/OpenManus-RL.