# Which Direction to Choose? An Analysis on the Representation Power of Self-Supervised ViTs in Downstream Tasks

arXiv: https://arxiv.org/abs/2509.15272 · HTML: https://mlsanity.com/paper/2509.15272/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 45 of its 53 references that are in the map.
- First posted on arXiv 18 Sep 2025.

## Builds on (most cited first, 30 of 45)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [A Simple Framework for Contrastive Learning of Visual Representations](https://mlsanity.com/paper/2002.05709/) (2020)
- [ImageNet Large Scale Visual Recognition Challenge](https://mlsanity.com/paper/1409.0575/) (2014)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [Training data-efficient image transformers & distillation through attention](https://mlsanity.com/paper/2012.12877/) (2020)
- [BEiT: BERT Pre-Training of Image Transformers](https://mlsanity.com/paper/2106.08254/) (2021)
- [Improved Baselines with Momentum Contrastive Learning](https://mlsanity.com/paper/2003.04297/) (2020)
- [Describing Textures in the Wild](https://mlsanity.com/paper/1311.3618/) (2013)
- [An Empirical Study of Training Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.02057/) (2021)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data](https://mlsanity.com/paper/2401.10891/) (2024)
- [CoCa: Contrastive Captioners are Image-Text Foundation Models](https://mlsanity.com/paper/2205.01917/) (2022)
- [SimMIM: A Simple Framework for Masked Image Modeling](https://mlsanity.com/paper/2111.09886/) (2021)
- [Transformers in Vision: A Survey](https://mlsanity.com/paper/2101.01169/) (2021)
- [PaLI: A Jointly-Scaled Multilingual Language-Image Model](https://mlsanity.com/paper/2209.06794/) (2022)
- [Network Dissection: Quantifying Interpretability of Deep Visual Representations](https://mlsanity.com/paper/1704.05796/) (2017)
- [Do Vision Transformers See Like Convolutional Neural Networks?](https://mlsanity.com/paper/2108.08810/) (2021)
- [A Survey on Visual Transformer](https://mlsanity.com/paper/2012.12556/) (2020)
- [Dense Contrastive Learning for Self-Supervised Visual Pre-Training](https://mlsanity.com/paper/2011.09157/) (2020)
- [Image as a Foreign Language: BEiT Pretraining for All Vision and Vision-Language Tasks](https://mlsanity.com/paper/2208.10442/) (2022)
- [Intriguing Properties of Vision Transformers](https://mlsanity.com/paper/2105.10497/) (2021)
- [Deep ViT Features as Dense Visual Descriptors](https://mlsanity.com/paper/2112.05814/) (2021)
- [BEiT v2: Masked Image Modeling with Vector-Quantized Visual Tokenizers](https://mlsanity.com/paper/2208.06366/) (2022)
- [Unsupervised Semantic Segmentation by Distilling Feature Correspondences](https://mlsanity.com/paper/2203.08414/) (2022)
- [Detect What You Can: Detecting and Representing Objects using Holistic Models and Body Parts](https://mlsanity.com/paper/1406.2031/) (2014)
- [Localizing Objects with Self-Supervised Transformers and no Labels](https://mlsanity.com/paper/2109.14279/) (2021)
- [EfficientSAM: Leveraged Masked Image Pretraining for Efficient Segment Anything](https://mlsanity.com/paper/2312.00863/) (2023)

## Abstract (arXiv)

Self-Supervised Learning (SSL) for Vision Transformers (ViTs) has recently demonstrated considerable potential as a pre-training strategy for a variety of computer vision tasks, including image classification and segmentation, both in standard and few-shot downstream contexts. Two pre-training objectives dominate the landscape of SSL techniques: Contrastive Learning and Masked Image Modeling. Features (or tokens) extracted from the final transformer attention block -- specifically, the keys, queries, and values -- as well as features obtained after the final block's feed-forward layer, have become a common foundation for addressing downstream tasks. However, in many existing approaches, these pre-trained ViT features are further processed through additional transformation layers, often involving lightweight heads or combined with distillation, to achieve superior task performance. Although such methods can improve task outcomes, to the best of our knowledge, a comprehensive analysis of the intrinsic representation capabilities of unaltered ViT features has yet to be conducted. This study aims to bridge this gap by systematically evaluating the use of these unmodified features across image classification and segmentation tasks, in both standard and few-shot contexts. The classification and segmentation rules that we use are either hyperplane based (as in logistic regression) or cosine-similarity based, both of which rely on the presence of interpretable directions in the ViT's latent space. Based on the previous rules and without the use of additional feature transformations, we conduct an analysis across token types, tasks, and pre-trained ViT models. This study provides insights into the optimal choice for token type and decision rule based on the task, context, and the pre-training objective, while reporting detailed findings on two widely-used datasets.