# How Emotionally Stable is ALBERT? Testing Robustness with Stochastic Weight Averaging on a Sentiment Analysis Task

arXiv: https://arxiv.org/abs/2111.09612 · HTML: https://mlsanity.com/paper/2111.09612/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 30 references that are in the map.
- First posted on arXiv 18 Nov 2021.

## Builds on (most cited first, 19 of 19)

- [GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding](https://mlsanity.com/paper/1804.07461/) (2018)
- [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://mlsanity.com/paper/1909.11942/) (2019)
- [Averaging Weights Leads to Wider Optima and Better Generalization](https://mlsanity.com/paper/1803.05407/) (2018)
- [Beyond Accuracy: Behavioral Testing of NLP models with CheckList](https://mlsanity.com/paper/2005.04118/) (2020)
- [Underspecification Presents Challenges for Credibility in Modern Machine Learning](https://mlsanity.com/paper/2011.03395/) (2020)
- [Polyjuice: Generating Counterfactuals for Explaining, Evaluating, and Improving Models](https://mlsanity.com/paper/2101.00288/) (2021)
- [Robustness Gym: Unifying the NLP Evaluation Landscape](https://mlsanity.com/paper/2101.04840/) (2021)
- [The MultiBERTs: BERT Reproductions for Robustness Analysis](https://mlsanity.com/paper/2106.16163/) (2021)
- [BERTs of a feather do not generalize together: Large variability in generalization across models with similar test set performance](https://mlsanity.com/paper/1911.02969/) (2019)
- [Memorization vs. Generalization: Quantifying Data Leakage in NLP Performance Evaluation](https://mlsanity.com/paper/2102.01818/) (2021)
- [The Spotlight: A General Method for Discovering Systematic Errors in Deep Learning Models](https://mlsanity.com/paper/2107.00758/) (2021)
- [Are Larger Pretrained Language Models Uniformly Better? Comparing Performance at the Instance Level](https://mlsanity.com/paper/2105.06020/) (2021)
- [Robustness Testing of Language Understanding in Task-Oriented Dialog](https://mlsanity.com/paper/2012.15262/) (2020)
- [On Model Stability as a Function of Random Seed](https://mlsanity.com/paper/1909.10447/) (2019)
- [Noise Stability Regularization for Improving BERT Fine-tuning](https://mlsanity.com/paper/2107.04835/) (2021)
- [Improving BERT Fine-Tuning via Self-Ensemble and Self-Distillation](https://mlsanity.com/paper/2002.10345/) (2020)
- [BiasFinder: Metamorphic Test Generation to Uncover Bias for Sentiment Analysis Systems](https://mlsanity.com/paper/2102.01859/) (2021)
- [Agree to Disagree: When Deep Learning Models With Identical Architectures Produce Distinct Explanations](https://mlsanity.com/paper/2105.06791/) (2021)
- [On the Impact of Random Seeds on the Fairness of Clinical Classifiers](https://mlsanity.com/paper/2104.06338/) (2021)

## Built on by (most cited first, 5 of 5)

- [A Survey on Stability of Learning with Limited Labelled Data and its Sensitivity to the Effects of Randomness](https://mlsanity.com/paper/2312.01082/) (2023)
- [Measuring the Instability of Fine-Tuning](https://mlsanity.com/paper/2302.07778/) (2023)
- [Uncertainty-Aware Natural Language Inference with Stochastic Weight Averaging](https://mlsanity.com/paper/2304.04726/) (2023)
- [Fighting Randomness with Randomness: Mitigating Optimisation Instability of Fine-Tuning using Delayed Ensemble and Noisy Interpolation](https://mlsanity.com/paper/2406.12471/) (2024)
- [You Shall Know a Tool by the Traces it Leaves: The Predictability of Sentiment Analysis Tools](https://mlsanity.com/paper/2410.14626/) (2024)

## Abstract (arXiv)

Despite their success, modern language models are fragile. Even small changes in their training pipeline can lead to unexpected results. We study this phenomenon by examining the robustness of ALBERT (arXiv:1909.11942) in combination with Stochastic Weight Averaging (SWA) (arXiv:1803.05407) -- a cheap way of ensembling -- on a sentiment analysis task (SST-2). In particular, we analyze SWA's stability via CheckList criteria (arXiv:2005.04118), examining the agreement on errors made by models differing only in their random seed. We hypothesize that SWA is more stable because it ensembles model snapshots taken along the gradient descent trajectory. We quantify stability by comparing the models' mistakes with Fleiss' Kappa (Fleiss, 1971) and overlap ratio scores. We find that SWA reduces error rates in general; yet the models still suffer from their own distinct biases (according to CheckList).