What capabilities does ALFRED test?

ALFRED evaluates embodied, image understanding, planning, instruction following.

How can a model improve its ALFRED score?

Tools linked to ALFRED on Sophon include ALFWorld - RL environments, datasets, and scaffolds that target this eval.

ALFRED

Active

3D-simulated household tasks driven by language instructions and egocentric video - the visual sibling of ALFWorld.

Open

Publisher: University of California, Berkeley
Capabilities: Embodied Image Understanding Planning Instruction Following
Domain: robotics
Format: Custom
Size: 25743 tasks
License: MIT
Published: Dec 2019
Notable for: Benchmark for evaluating embodied, image understanding and planning in the robotics domain.
Canonical: askforalfred.com
Also on: github.com/askforalfred/alfred

Cite

Notes

Only stored in your browser.

Related tools

View all

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

ALFWorld

MIT CSAIL

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

FAQ

What is ALFRED?: 3D-simulated household tasks driven by language instructions and egocentric video - the visual sibling of ALFWorld.
What capabilities does ALFRED test?: ALFRED evaluates embodied, image understanding, planning, instruction following.
How can a model improve its ALFRED score?: Tools linked to ALFRED on Sophon include ALFWorld - RL environments, datasets, and scaffolds that target this eval.
What license is ALFRED under?: ALFRED is available under MIT.