scienceworld-act

To execute actions in a ScienceWorld simulated environment, send text commands like "look" or "take watering can" to perform object manipulation and environmental exploration, receiving observations and rewards after each action.

An AI agent simulation environment for interactive scientific experimentation allows agents to perform deterministic tasks, manipulate virtual objects, and conduct experiments to achieve scientific goals like growing a plant within a simulated text-based world.

Yes, you need an active ScienceWorld session managed by an executive node. This setup eliminates the need for explicit session IDs, ensuring seamless integration when executing scientific experimentation tasks.

You get observations and rewards after performing tasks in a science simulation by sending text commands to the environment. The system returns detailed feedback, rewards, and task completion status after each action is executed.

Yes, you can use this skill to conduct virtual lab experiments and problem-solving. An agent can explore the virtual lab, manipulate objects, and run experiments to achieve scientific goals within the ScienceWorld environment.

No, the ScienceWorld environment interaction skill does not require managing explicit session IDs. It integrates with an active session managed by an executive node, abstracting away session management for seamless agent interaction.