Challenge: Existing benchmarks have exposed patterns and may not truly assess generalization ability of Large Language Models (LLMs).
Approach: They propose a “Generalization Stress Test” to assess Large Language Models’ generalization ability under slight and controlled perturbations, including option length, problem types, and irrelevant noun replacements.
Outcome: The proposed test shows that LLMs exhibit severe accuracy drops and unexpected biases when faced with minor but content-preserving modifications.

Similar Papers

Unveiling the Generalization Power of Fine-Tuned Large Language Models (2024.naacl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) have demonstrated exceptional multitasking abilities, but the comprehensive effects of fine-tuning on the LLMs’ generalization ability are not fully understood.
Approach: They conduct extensive experiments across five distinct language tasks on different datasets to investigate whether fine-tuning affects the generalization ability intrinsic to LLMs.
Outcome: The proposed model can generalize to different domains and tasks by integrating the in-context learning strategy during fine-tuning on generation tasks.
Methods for Estimating and Improving Robustness of Language Models (2022.naacl-srw)

Copied to clipboard

Challenge: Large language models suffer from weak generalisation ability due to shallow textual relations over full semantic complexity of the problem.
Approach: They propose to incorporate some of these measures into training objectives to enhance distributional robustness of LLMs.
Outcome: The proposed models outperform human models on complex tasks and outperformed other models on deep networks.
Do LLMs Overcome Shortcut Learning? An Evaluation of Shortcut Challenges in Large Language Models (2024.emnlp-main)

Copied to clipboard

Challenge: Large Language Models (LLMs) have shown remarkable capabilities in various tasks, but may rely on dataset biases as shortcuts for prediction.
Approach: They propose to use a test suite to evaluate the impact of shortcuts on LLMs' performance.
Outcome: The proposed test suite incorporates six shortcut types, five evaluation metrics, and four prompting strategies.
Exploring Compositional Generalization of Large Language Models (2024.naacl-srw)

Copied to clipboard

Challenge: a recent study has found that large language models can generalize compositional instructions from simple instructions to complex ones.
Approach: They study the generalization ability of large language models with respect to compositional instructions . they first construct a dataset with the help of ChatGPT guided by the self-instruct technique .
Outcome: The proposed model can generalize from simple instructions to more intricate ones, the authors show . their results show that training LLMs on higher-order compositional instructions improves performance on lower-order ones, but not on higher order ones.
Which Word Orders Facilitate Length Generalization in LMs? An Investigation with GCG-Based Artificial Languages (2025.emnlp-main)

Copied to clipboard

Challenge: Whether language models have inductive biases favoring typologically frequent grammatical properties over rare, implausible ones has been investigated, typically using artificial languages (ALs).
Approach: They extend their context-free AL formalization by adopting Generalized Categorial Grammar (GCG) . they also examine the generalization ability of LMs to process unseen longer test sentences .
Outcome: The proposed models better capture features of natural languages and can process unseen longer test sentences.
Quantifying Generalizations: Exploring the Divide Between Human and LLMs’ Sensitivity to Quantification (2024.acl-long)

Copied to clipboard

Challenge: Generics are expressions used to communicate abstractions about categories . they allow for exceptions, and they are a powerful way to express knowledge about the world .
Approach: They examine how large language models interpret generics to understand their meanings . they find that the presence of a generic sentence as context influences quantifiers based on the generalization .
Outcome: The proposed models do not exhibit a strong sensitivity to quantification, the study finds . the results suggest that the presence of a generic sentence as context influences quantifiers .
How Abstract Is Linguistic Generalization in Large Language Models? Experiments with Argument Structure (2023.tacl-1)

Copied to clipboard

Challenge: Competent speakers of a language know how likely a word w is to appear in a specific context .
Approach: They use transformer-based large language models to generalize a novel noun argument . they show a bias to generalise based on linear order, instead of a linear order .
Outcome: The proposed models perform well in generalizing the distribution of a novel noun argument between related contexts that were seen during pre-training.
Revisiting Generalization Across Difficulty Levels: It’s Not So Easy (2026.eacl-long)

Copied to clipboard

Challenge: Existing research is mixed regarding whether training on easier or harder data leads to better results.
Approach: They examine how well large language models generalize across different task difficulties by using a large dataset and a well-established difficulty metric.
Outcome: The results show that training on hard data can't achieve consistent improvements across the full range of difficulties.
Detecting Edit Failures In Large Language Models: An Improved Specificity Benchmark (2023.findings-acl)

Copied to clipboard

Challenge: Recent model editing techniques can introduce large unwanted side effects, a new study shows . existing specificity benchmarks do not detect these unwanted side-effects . a recent study shows that model edits can cause significant performance drop .
Approach: They extend existing CounterFact benchmark to include a dynamic component and propose a new benchmark to evaluate model editing techniques.
Outcome: The proposed benchmark improves existing benchmarks for specificity and avoids unwanted side effects.
Large Language Models Sensitivity to The Order of Options in Multiple-Choice Questions (2024.findings-naacl)

Copied to clipboard

Challenge: Large Language Models have demonstrated impressive capabilities in various NLP tasks, but previous studies have shown they are sensitive to prompt wording and few-shot demonstrations and their order.
Approach: They focus on LLMs robustness on multiple-choice questions . they find a performance gap of 13% to 85% when options are reordered .
Outcome: The proposed model outperforms supervised models on multiple choice questions even outperforming humans.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations