0

ALFWorld

Active

Embodied household-task benchmark that aligns TextWorld text commands with ALFRED 3D scenes, testing whether agents can transfer from abstract text policies to grounded execution.

Open
Publisher
Microsoft
Capabilities
EmbodiedPlanningTool CallingCommon SenseMulti Turn Dialog
Domain
agentic
Format
Custom
Size
3,553 tasks across 6 task types tasks
License
MIT
Published
Oct 2020
Notable for
Benchmark for evaluating embodied, planning and tool calling in the agentic domain.
Canonical
alfworld.github.io

Cite

Notes

Only stored in your browser.

Related tools

1
View all

Implementations, trainers, datasets and scaffolds linked to this eval.

ALFWorld

MIT CSAIL

Official

Aligned text-and-3D embodied environment - agents learn household tasks (pick & place, heat, cool, clean) as both TextWorld games and visually-rendered ALFRED scenes.

Trains towardRL EnvPlanningEmbodiedInstruction Following

Papers

2

ALFWorld: Aligning Text and Embodied Environments for Interactive Learning

ICLR · 2021

A text-game version of the ALFRED household task benchmark that lets agents practice in TextWorld, then transfer to embodied AI2-THOR.

introduces

ALFWorld: Aligning Text and Embodied Environments for Interactive Learning

ICLR · 2021

A text-game version of the ALFRED household task benchmark that lets agents practice in TextWorld, then transfer to embodied AI2-THOR.

FAQ

What is ALFWorld?
Embodied household-task benchmark that aligns TextWorld text commands with ALFRED 3D scenes, testing whether agents can transfer from abstract text policies to grounded execution.
What capabilities does ALFWorld test?
ALFWorld evaluates embodied, planning, tool calling, common sense, multi turn dialog.
How can a model improve its ALFWorld score?
Tools linked to ALFWorld on Sophon include ALFWorld - RL environments, datasets, and scaffolds that target this eval.
What license is ALFWorld under?
ALFWorld is available under MIT.