AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

URLB: Unsupervised Reinforcement Learning Benchmark

Michael Laskin, Denis Yarats, Hao Liu, Kimin Lee, Albert Zhan, Kevin Lu, Catherine Cang, Lerrel Pinto, Pieter Abbeel

NeurIPS 2021 · Reinforcement learning › Visual and representation RL · approach: Unsupervised skill discovery · 11 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
unsupervised pre-training up to 2M steps then 100k fine-tune vs best methodsall eight baselines · URLB (12 tasks)normalized return: 40-70% normalized return vs ~100% neededMixed
pre-train with intrinsic rewards instead of random initialization (states) (ablation)DDPG · URLB statesextrinsic returnNo clear effect
pre-train with intrinsic rewards instead of random initialization (pixels) (ablation)DrQ-v2 · URLB pixelsextrinsic return: Walker Stand 212 -> 552 (Disagreement, 2M)Improved
switch observations from states to pixelsexploration algorithms · URLBnormalized return: lose 20-50%Hurt
competence-based vs knowledge/data-based intrinsic rewardsunsupervised RL · URLBaggregate normalized returnHurt
increase pre-training steps from 100k to 2M (ablation)APS · Walker Flip, statesextrinsic return: 638 -> 461Hurt
increase pre-training steps from 100k to 2M (ablation)Disagreement · Quadruped Walk, statesextrinsic return: 265 -> 721Improved
increase pre-training steps (ablation)unsupervised RL algorithms · URLB, 18 experimentsfine-tuning performance: 9 of 18 do not improve or degradeMixed

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