AI Paper Atlas

Research map / Reinforcement learning / Imitation learning

Improving Policy Learning via Language Dynamics Distillation

Victor Zhong, Jesse Mu, Luke Zettlemoyer, Edward Grefenstette, Tim Rocktäschel

NeurIPS 2022 · Reinforcement learning › Imitation learning · approach: Imitation combined with RL · 5 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
Pretrain and distill language-grounded dynamics representations from unlabeled demonstrationsPolicy learning on SILG tasks · NetHack, ALFWorld, RTFM, Messenger, Touchdowntask performanceImproved
Include language descriptions in dynamics-modeling observations (ablation)LDD · NetHackpolicy performanceImproved
Include language descriptions in dynamics-modeling observations (ablation)LDD · SymTouchdownpolicy performanceImproved
Train the dynamics model on expert demonstrations instead of non-expert rollouts (ablation)LDD · SILG environmentsdynamics-model accuracyImproved
Pretrain to initialize the policy learner (ablation)LDD · Messengerpolicy performanceImproved
Distill the pretrained dynamics representations during policy learning (ablation)LDD · Touchdownevaluation win rateImproved
Use dynamics-based reward shaping instead of LDDPolicy learning · SILG environmentstask performanceHurt
Use VAE pretraining instead of LDD dynamics pretrainingPolicy learning · SILG environmentstask performanceHurt
Use inverse reinforcement learning instead of LDDPolicy learning · SILG environmentstask performanceHurt
Remove the initialization step (ablation)LDD · Messengerpolicy performanceHurt
Remove the distillation step (ablation)LDD · Touchdownevaluation win rateHurt
Remove in-game messages and character-state descriptions (ablation)LDD dynamics-modeling observations · NetHackpolicy performanceHurt

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