AI Paper Atlas

Research map / Video / Video generation

Learning Universal Policies via Text-Guided Video Generation

Yilun Du, Sherry Yang, Bo Dai, Hanjun Dai, Ofir Nachum, Joshua B. Tenenbaum, Dale Schuurmans, Pieter Abbeel

NeurIPS 2023 · Video › Video generation · approach: Compositional text-to-video diffusion · 29 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
condition video model on first observation frame (train-time) (ablation)UniPi · Combinatorial (seen) Placeaccuracy: 13.2 -> 52.4Improved
condition video model on first observation frame (train-time) (ablation)UniPi · Combinatorial (seen) Relationaccuracy: 12.4 -> 34.7Improved
add frame consistency by tiling observed frame (ablation)UniPi · Combinatorial (seen) Relationaccuracy: 34.7 -> 39.4Improved
add frame consistency by tiling observed frame (ablation)UniPi · Combinatorial (seen) Placeaccuracy: 52.4 -> 53.2No clear effect
add temporal hierarchy (temporal super-resolution) (ablation)UniPi · Combinatorial (seen) Relationaccuracy: 39.4 -> 53.2Improved
add temporal hierarchy (temporal super-resolution) (ablation)UniPi · Combinatorial (seen) Placeaccuracy: 53.2 -> 59.1Improved
pretrain on internet-scale video-text data (ablation)UniPi · Bridge real robot videoFVD: 288.02 -> 264.66Improved
pretrain on internet-scale video-text data (ablation)UniPi · Bridge real robot videoFID: 17.75 -> 14.54Improved
pretrain on internet-scale video-text data (ablation)UniPi · Bridge real robot videoCLIP score: 24.43 -> 24.54No clear effect
pretrain on internet-scale video-text data (ablation)UniPi · Bridge real robot videosuccess classifier accuracy: 72.6% -> 77.1%Improved
predict future video frames then extract actions instead of diffusing actions directly (Diffuser)diffusion planner · Combinatorial (novel) Relationtask completion accuracy: 9.6 -> 46.1Improved

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