AI Paper Atlas

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

Episodic Transformer for Vision-and-Language Navigation

Alexander Pashevich, Cordelia Schmid, Chen Sun

ICCV 2021 · Vision-language and multimodal › Robot action and navigation · approach: Pretraining and data augmentation for VLN · 160 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
replace LSTM with episodic transformerLSTM agent · ALFRED validation seentask success (%): 23.2 -> 33.8Improved
replace LSTM with episodic transformerLSTM agent · ALFRED validation unseentask success (%): 2.4 -> 3.2Improved
replace LSTM with episodic transformerLSTM agent · ALFRED validation sub-goals seensub-goal success (%): 75.5 -> 77.3No clear effect
give the LSTM the E.T. language encoder (ablation)LSTM agent · ALFRED validation seentask success (%): 23.2 -> 27.8Improved
add access to the full history of visual frames instead of the last frame (ablation)E.T. · ALFRED validation seentask success (%): 28.9 -> 33.8Improved
add access to more than the last action (all actions instead of 1) (ablation)E.T. · ALFRED validation seentask success (%): 33.8 -> 27.1Hurt
add the last action as input (versus none) (ablation)E.T. · ALFRED validation seentask success (%): 23.7 -> 33.8Improved
use 2 transformer blocks instead of 1 (ablation)E.T. · ALFRED validation seentask success (%): 25.0 -> 33.8Improved
use 3 blocks instead of 2 (ablation)E.T. · ALFRED validation seentask success (%): 33.8 -> 28.6Hurt
add joint training with subgoal-action synthetic instructions (ablation)E.T. · ALFRED validation seentask success (%): 33.8 -> 38.5Improved
add joint training with expert-frame instructions (ablation)E.T. · ALFRED validation seentask success (%): 33.8 -> 28.5Hurt
add BERT embeddings for language pretraining (ablation)E.T. · ALFRED validation seentask success (%): 33.8 -> 32.3Hurt

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