pith. machine review for the scientific record. sign in

arxiv: 1902.01007 · v4 · submitted 2019-02-04 · 💻 cs.CL

Recognition: unknown

Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inference

Ellie Pavlick, R. Thomas McCoy, Tal Linzen

classification 💻 cs.CL
keywords heuristicsheuristichansmodelsadoptedinferencelanguagenatural
0
0 comments X
read the original abstract

A machine learning system can score well on a given test set by relying on heuristics that are effective for frequent example types but break down in more challenging cases. We study this issue within natural language inference (NLI), the task of determining whether one sentence entails another. We hypothesize that statistical NLI models may adopt three fallible syntactic heuristics: the lexical overlap heuristic, the subsequence heuristic, and the constituent heuristic. To determine whether models have adopted these heuristics, we introduce a controlled evaluation set called HANS (Heuristic Analysis for NLI Systems), which contains many examples where the heuristics fail. We find that models trained on MNLI, including BERT, a state-of-the-art model, perform very poorly on HANS, suggesting that they have indeed adopted these heuristics. We conclude that there is substantial room for improvement in NLI systems, and that the HANS dataset can motivate and measure progress in this area

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Language Models are Few-Shot Learners

    cs.CL 2020-05 accept novelty 8.0

    GPT-3 shows that scaling an autoregressive language model to 175 billion parameters enables strong few-shot performance across diverse NLP tasks via in-context prompting without fine-tuning.

  2. Multitask Prompted Training Enables Zero-Shot Task Generalization

    cs.LG 2021-10 conditional novelty 7.0

    Multitask fine-tuning of an encoder-decoder model on prompted datasets produces zero-shot generalization that often beats models up to 16 times larger on standard benchmarks.

  3. BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions

    cs.CL 2019-05 accept novelty 7.0

    BoolQ introduces naturally occurring yes/no questions as a challenging benchmark where BERT fine-tuned on MultiNLI reaches 80.4% accuracy against 90% human performance.

  4. Decaf: Improving Neural Decompilation with Automatic Feedback and Search

    cs.SE 2026-05 unverdicted novelty 6.0

    Decaf uses compiler feedback and search to improve neural decompilation, boosting semantic success rate from 26.0% to 83.9% on ExeBench Real -O2 split.

  5. Deciphering Shortcut Learning from an Evolutionary Game Theory Perspective

    cs.AI 2026-05 unverdicted novelty 6.0

    Evolutionary game theory shows gradient descent and stochastic gradient descent drive neural networks to distinct stable states favoring shortcut or core subnetworks, with data and optimization noise shaping shortcut ...