scienceworld-conditional-focus-executor

To execute a conditional focus action in ScienceWorld, you use a skill that automates focusing on a specific object, such as a colored box, contingent on prior measurement results. It requires a clear input indicating the satisfied condition.

Conditional focus logic for agent tasks in ScienceWorld is a mechanism that enables an agent to interact with specific objects based on prior experimental results, automating the final step in conditional measurement tasks.

You make an agent focus on a specific object by providing a pre-determined conditional outcome as input. For example, after measuring temperature, the agent focuses on the red box if the temperature is high, or the blue box if it is low.

Yes, conditional focus execution in ScienceWorld is input-driven and relies entirely on a pre-determined conditional outcome from a prior measurement to decide which designated object to focus on.

Yes, you can use conditional focus execution for any ScienceWorld scenario requiring an agent to interact with specific objects contingent upon experimental results, provided the prior measurement outcome is clearly indicated as input.

A conditional focus action fails to execute on the correct object in ScienceWorld if the input does not clearly indicate the satisfied condition; the skill requires an explicit pre-determined conditional outcome to perform the correct focus action.