Counterfactual Invariance to Spurious Correlations in Text Classification

Authors: Victor Veitch, Alexander D'Amour, Steve Yadlowsky, Jacob Eisenstein

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental This theory is supported by empirical results on text classification.
Researcher Affiliation Collaboration Victor Veitch1,2, Alexander D Amour1, Steve Yadlowsky1, and Jacob Eisenstein1 1Google Research 2University of Chicago
Pseudocode No No structured pseudocode or algorithm blocks were found.
Open Source Code No No explicit statement or link providing access to the source code for the methodology was found. The paper mentions 'See supplement for experimental details' but does not confirm code availability.
Open Datasets Yes We build the experimental datasets using Amazon reviews from the product category Clothing, Shoes, and Jewelry [NLM19]. For an additional test on naturally-occurring confounds, we use the multigenre natural language inference (MNLI) dataset [WNB18].
Dataset Splits No The paper mentions training and test data, but does not provide specific train/validation/test split percentages or sample counts in the main text.
Hardware Specification No No specific hardware details (e.g., CPU, GPU, or TPU models) used for running experiments were provided.
Software Dependencies No The paper mentions using BERT but does not provide specific version numbers for software dependencies or libraries.
Experiment Setup No The paper mentions using BERT and training with a Cross Entropy + λ Regularizer, varying λ. However, specific hyperparameter values (e.g., learning rate, batch size, number of epochs) are not provided in the main text, with details deferred to the supplement.