Large Pre-Training Datasets Don't Always Guarantee Robustness after Fine-Tuning

• URL: http://arxiv.org/abs/2410.21582v3
• Date: Fri, 26 Sep 2025 17:57:05 GMT
• Title: Large Pre-Training Datasets Don't Always Guarantee Robustness after Fine-Tuning
• Authors: Jaedong Hwang, Brian Cheung, Zhang-Wei Hong, Akhilan Boopathy, Pulkit Agrawal, Ila Fiete,
• Abstract summary: We propose the Robustness Inheritance Benchmark (ImageNet-RIB) to assess robustness preservation in fine-tuned models.<n>We find that fine-tuning reduces robustness across pretrained models.<n>Models pretrained on the largest and most diverse datasets exhibit both larger robustness losses and lower absolute robustness after fine-tuning on small datasets.
• Score: 29.69990405081772
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large-scale pretrained models are widely leveraged as foundations for learning new specialized tasks via fine-tuning, with the goal of maintaining the general performance of the model while allowing it to gain new skills. A valuable goal for all such models is robustness: the ability to perform well on out-of-distribution (OOD) tasks. We assess whether fine-tuning preserves the overall robustness of the pretrained model, and observed that models pretrained on large datasets exhibited strong catastrophic forgetting and loss of OOD generalization. To systematically assess robustness preservation in fine-tuned models, we propose the Robustness Inheritance Benchmark (ImageNet-RIB). The benchmark, which can be applied to any pretrained model, consists of a set of related but distinct OOD (downstream) tasks and involves fine-tuning on one of the OOD tasks in the set then testing on the rest. We find that though continual learning methods help, fine-tuning reduces robustness across pretrained models. Surprisingly, models pretrained on the largest and most diverse datasets (e.g., LAION-2B) exhibit both larger robustness losses and lower absolute robustness after fine-tuning on small datasets, relative to models pretrained on smaller datasets. These findings suggest that starting with the strongest foundation model is not necessarily the best approach for performance on specialist tasks. https://jd730.github.io/projects/ImageNet-RIB

Related papers

• Feed Two Birds with One Scone: Exploiting Function-Space Regularization for Both OOD Robustness and ID Fine-Tuning Performance [72.57668440744301]
  We propose a novel regularization that constrains the distance of fine-tuning and pre-trained model in the function space with simulated OOD samples.<n>Our approach could consistently improve both downstream task ID fine-tuning performance and OOD robustness across a variety of CLIP backbones.
  arXiv  Detail & Related papers  (2025-08-31T12:14:34Z)
• SCAN: Bootstrapping Contrastive Pre-training for Data Efficiency [10.555957282859]
  This paper introduces a novel dynamic bootstrapping dataset pruning method. It involves pruning data preparation followed by dataset mutation operations, both of which undergo iterative and dynamic updates. We individually pre-train seven CLIP models on two large-scale image-text pair datasets, and two MoCo models on the ImageNet dataset, resulting in a total of 16 pre-trained models.
  arXiv  Detail & Related papers  (2024-11-14T01:53:17Z)
• Automatic Pruning of Fine-tuning Datasets for Transformer-based Language Models [13.340191056212692]
  We propose an automatic dataset pruning method for the training set of fine-tuning tasks. Our method provides multiple subsets for use in dataset pruning. Experiments on 5 downstream tasks and 2 language models show that, on average, fine-tuning on the winning ticket subsets results in a $0.1 %$ increase in the evaluation performance of the model.
  arXiv  Detail & Related papers  (2024-07-11T22:46:18Z)
• Feature Protection For Out-of-distribution Generalization [24.072876186625855]
  We show that protecting pre-trained features leads to a fine-tuned model more robust to generalization. We show that protecting pre-trained features leads to a fine-tuned model more robust to OOD generalization.
  arXiv  Detail & Related papers  (2024-05-25T03:00:06Z)
• Model Stock: All we need is just a few fine-tuned models [34.449901046895185]
  This paper introduces an efficient fine-tuning method for large pre-trained models, offering strong in-distribution (ID) and out-of-distribution (OOD) performance.<n>We employ significantly fewer models to achieve final weights yet yield superior accuracy.<n>We demonstrate the efficacy of Model Stock with fine-tuned models based upon pre-trained CLIP architectures.
  arXiv  Detail & Related papers  (2024-03-28T15:57:20Z)
• Learning with Noisy Foundation Models [95.50968225050012]
  This paper is the first work to comprehensively understand and analyze the nature of noise in pre-training datasets. We propose a tuning method (NMTune) to affine the feature space to mitigate the malignant effect of noise and improve generalization.
  arXiv  Detail & Related papers  (2024-03-11T16:22:41Z)
• LEVI: Generalizable Fine-tuning via Layer-wise Ensemble of Different Views [28.081794908107604]
  Fine-tuning is used to leverage the power of pre-trained foundation models in new downstream tasks. Recent studies have observed challenges in the generalization of fine-tuned models to unseen distributions. We propose a novel generalizable fine-tuning method LEVI, where the pre-trained model is adaptively ensembled layer-wise with a small task-specific model.
  arXiv  Detail & Related papers  (2024-02-07T08:16:40Z)
• Not All Data Matters: An End-to-End Adaptive Dataset Pruning Framework for Enhancing Model Performance and Efficiency [9.460023981858319]
  We propose an end-to-end Adaptive DAtaset PRUNing framework called AdaPruner. AdaPruner iteratively prunes redundant samples to an expected pruning ratio. It can still significantly enhance model performance even after pruning up to 10-30% of the training data.
  arXiv  Detail & Related papers  (2023-12-09T16:01:21Z)
• A Simple and Efficient Baseline for Data Attribution on Images [107.12337511216228]
  Current state-of-the-art approaches require a large ensemble of as many as 300,000 models to accurately attribute model predictions. In this work, we focus on a minimalist baseline, utilizing the feature space of a backbone pretrained via self-supervised learning to perform data attribution. Our method is model-agnostic and scales easily to large datasets.
  arXiv  Detail & Related papers  (2023-11-03T17:29:46Z)
• Fantastic Gains and Where to Find Them: On the Existence and Prospect of General Knowledge Transfer between Any Pretrained Model [74.62272538148245]
  We show that for arbitrary pairings of pretrained models, one model extracts significant data context unavailable in the other. We investigate if it is possible to transfer such "complementary" knowledge from one model to another without performance degradation.
  arXiv  Detail & Related papers  (2023-10-26T17:59:46Z)
• Domain Generalization Using Large Pretrained Models with Mixture-of-Adapters [33.401355417911084]
  This study is on leveraging the knowledge of large pretrained models to improve handling of OOD scenarios and tackle domain generalization problems.<n>We employ parameter-efficient fine-tuning (PEFT) techniques to effectively preserve OOD robustness while working with large models.<n>Our experiments and analysis confirm that the most effective approaches involve ensembling diverse models and increasing the scale of pretraining.
  arXiv  Detail & Related papers  (2023-10-17T07:01:24Z)
• Alleviating the Effect of Data Imbalance on Adversarial Training [26.36714114672729]
  We study adversarial training on datasets that obey the long-tailed distribution. We propose a new adversarial training framework -- Re-balancing Adversarial Training (REAT)
  arXiv  Detail & Related papers  (2023-07-14T07:01:48Z)
• TWINS: A Fine-Tuning Framework for Improved Transferability of Adversarial Robustness and Generalization [89.54947228958494]
  This paper focuses on the fine-tuning of an adversarially pre-trained model in various classification tasks. We propose a novel statistics-based approach, Two-WIng NormliSation (TWINS) fine-tuning framework. TWINS is shown to be effective on a wide range of image classification datasets in terms of both generalization and robustness.
  arXiv  Detail & Related papers  (2023-03-20T14:12:55Z)
• Effective Robustness against Natural Distribution Shifts for Models with Different Training Data [113.21868839569]
  "Effective robustness" measures the extra out-of-distribution robustness beyond what can be predicted from the in-distribution (ID) performance. We propose a new evaluation metric to evaluate and compare the effective robustness of models trained on different data.
  arXiv  Detail & Related papers  (2023-02-02T19:28:41Z)
• How to Train an Accurate and Efficient Object Detection Model on Any Dataset [0.0]
  We propose a dataset-agnostic template for object detection trainings. It consists of carefully chosen and pre-trained models together with a robust training pipeline for further training. Our solution works out-of-the-box and provides a strong baseline on a wide range of datasets.
  arXiv  Detail & Related papers  (2022-11-30T17:09:01Z)
• Are Sample-Efficient NLP Models More Robust? [90.54786862811183]
  We investigate the relationship between sample efficiency (amount of data needed to reach a given ID accuracy) and robustness (how models fare on OOD evaluation) We find that higher sample efficiency is only correlated with better average OOD robustness on some modeling interventions and tasks, but not others. These results suggest that general-purpose methods for improving sample efficiency are unlikely to yield universal OOD robustness improvements, since such improvements are highly dataset- and task-dependent.
  arXiv  Detail & Related papers  (2022-10-12T17:54:59Z)
• SynBench: Task-Agnostic Benchmarking of Pretrained Representations using Synthetic Data [78.21197488065177]
  Recent success in fine-tuning large models, that are pretrained on broad data at scale, on downstream tasks has led to a significant paradigm shift in deep learning. This paper proposes a new task-agnostic framework, textitSynBench, to measure the quality of pretrained representations using synthetic data.
  arXiv  Detail & Related papers  (2022-10-06T15:25:00Z)
• Self-Distillation for Further Pre-training of Transformers [83.84227016847096]
  We propose self-distillation as a regularization for a further pre-training stage. We empirically validate the efficacy of self-distillation on a variety of benchmark datasets for image and text classification tasks.
  arXiv  Detail & Related papers  (2022-09-30T02:25:12Z)
• Towards Inadequately Pre-trained Models in Transfer Learning [37.66278189011681]
  Better ImageNet pre-trained models have been demonstrated to have better transferability to downstream tasks. In this paper, we found that during the same pre-training process, models at middle epochs, which is inadequately pre-trained, can outperform fully trained models. Our discoveries suggest that, during pre-training, models tend to first learn spectral components corresponding to large singular values.
  arXiv  Detail & Related papers  (2022-03-09T12:15:55Z)
• Improved Fine-tuning by Leveraging Pre-training Data: Theory and Practice [52.11183787786718]
  Fine-tuning a pre-trained model on the target data is widely used in many deep learning applications. Recent studies have empirically shown that training from scratch has the final performance that is no worse than this pre-training strategy. We propose a novel selection strategy to select a subset from pre-training data to help improve the generalization on the target task.
  arXiv  Detail & Related papers  (2021-11-24T06:18:32Z)
• DSEE: Dually Sparsity-embedded Efficient Tuning of Pre-trained Language Models [152.29364079385635]
  As pre-trained models grow bigger, the fine-tuning process can be time-consuming and computationally expensive. We propose a framework for resource- and parameter-efficient fine-tuning by leveraging the sparsity prior in both weight updates and the final model weights. Our proposed framework, dubbed Dually Sparsity-Embedded Efficient Tuning (DSEE), aims to achieve two key objectives: (i) parameter efficient fine-tuning and (ii) resource-efficient inference.
  arXiv  Detail & Related papers  (2021-10-30T03:29:47Z)
• Comparing Test Sets with Item Response Theory [53.755064720563]
  We evaluate 29 datasets using predictions from 18 pretrained Transformer models on individual test examples. We find that Quoref, HellaSwag, and MC-TACO are best suited for distinguishing among state-of-the-art models. We also observe span selection task format, which is used for QA datasets like QAMR or SQuAD2.0, is effective in differentiating between strong and weak models.
  arXiv  Detail & Related papers  (2021-06-01T22:33:53Z)
• Efficient Conditional Pre-training for Transfer Learning [71.01129334495553]
  We propose efficient filtering methods to select relevant subsets from the pre-training dataset. We validate our techniques by pre-training on ImageNet in both the unsupervised and supervised settings. We improve standard ImageNet pre-training by 1-3% by tuning available models on our subsets and pre-training on a dataset filtered from a larger scale dataset.
  arXiv  Detail & Related papers  (2020-11-20T06:16:15Z)
• Deep Ensembles for Low-Data Transfer Learning [21.578470914935938]
  We study different ways of creating ensembles from pre-trained models. We show that the nature of pre-training itself is a performant source of diversity. We propose a practical algorithm that efficiently identifies a subset of pre-trained models for any downstream dataset.
  arXiv  Detail & Related papers  (2020-10-14T07:59:00Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.