AI Paper Atlas

Research map / Probabilistic and scientific modeling / Causal discovery

Causal Abstractions of Neural Networks

Atticus Geiger, Hanson Lu, Thomas F Icard, Christopher Potts

NeurIPS 2021 · Probabilistic and scientific modeling › Causal discovery · approach: Causal representation identifiability · 0 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 labeled-subphrase data augmentation to training (ablation)BERT-base · MQNLI hard generalization testaccuracy: 54.51 -> 88.50Improved
add dev-set comparison for augmentation (ablation)BERT-base · MQNLI hard generalization devaccuracy: 55.42 -> 88.25Improved
use BERT pretraining instead of no pretraining (ablation)BERT-base · MQNLI hard generalizationaccuracy: 49.1 -> 88.50Improved
use BERT instead of BiLSTMMQNLI classifier · MQNLI NP_Obj causal abstractionclique size (%): 0.7 -> 38.3Improved
remove N_Obj^P from the NP_Obj composition (ablation)NP_Obj causal model · MQNLI on BERTclique size (%): 38.3 -> 33.8Hurt
remove A_Obj^H from the NP_Obj composition (ablation)NP_Obj causal model · MQNLI on BERTclique size (%): 38.3 -> 15.7Hurt
remove N_Obj^H and N_Obj^P from the NP_Obj composition (ablation)NP_Obj causal model · MQNLI on BERTclique size (%): 38.3 -> 14.1Hurt
add N_Subj^P to the NP_Obj composition (ablation)NP_Obj causal model · MQNLI on BERTclique size (%): 38.3 -> 37.2Hurt
add Adj_Subj^P to the NP_Obj composition (ablation)NP_Obj causal model · MQNLI on BERTclique size (%): 38.3 -> 30.5Hurt
use interchange interventions instead of probing to locate representationsBERT analysis · MQNLI NP_Objdiscriminativeness of locationImproved
use selectivity instead of probe accuracyprobing · MQNLI NP_Objprobe resultNo clear effect

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