AI Paper Atlas

Research map / Vision-language and multimodal / Robot action and navigation

AndroidInTheWild: A Large-Scale Dataset For Android Device Control

Christopher Rawles, Alice Li, Daniel Rodriguez, Oriana Riva, Timothy P Lillicrap

NeurIPS 2023 · Vision-language and multimodal › Robot action and navigation · approach: Embodied multimodal LLM agents · 13 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add 2-step screen-action history to the BC agent (ablation)BC agent · AitW all splitspartial match score: 68.7 -> 73.1 (standard); 52.2 -> 59.7 (domain)Improved
use 5-shot CoT with action history instead of zero-shot (ablation)PaLM 2 agent · AitW standardpartial match score: 30.9 -> 39.6Improved
use 5-shot CoT with history instead of zero-shot (ablation)PaLM 2 agent · AitW unseen Android versionpartial match score: 31.6 -> 29.5No clear effect
use trained BC agent instead of prompted LLMdevice-control agent · AitW standardpartial match score: 39.6 -> 73.1Improved
evaluate on OOD splits vs standardBC-history · AitWpartial match score: 73.1 -> 59.7 (domain)Hurt
restrict LLM to feasible (element-based) actionsLLM-hist-5-CoT · AitW standardpartial match score: 39.6 -> 58Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy