ALFWorld

ALFWorld is a benchmark dataset that contains tasks for evaluating the performance of embodied agents in both textual and visual contexts.

Papers using ALFWorld (4)

Embodied Task Planning via Graph-Informed Action Generation with Large Language Models2026MCTS-EP: Empowering Embodied Planning With Online Preference Optimization2025GFlowVLM: Enhancing Multi-step Reasoning in Vision-Language Models with Generative Flow Networks2025Embodied Multi-Modal Agent trained by an LLM from a Parallel TextWorld2023 · 3 cites