# Lossless Adaptation of Pretrained Vision Models For Robotic Manipulation

arXiv: https://arxiv.org/abs/2304.06600 · HTML: https://mlsanity.com/paper/2304.06600/

## Key facts

- Cited by 29 papers in the map (data through 14 Oct 2026).
- Builds on 59 of its 64 references that are in the map.
- First posted on arXiv 13 Apr 2023.

## Builds on (most cited first, 30 of 59)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [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)
- [End-to-End Object Detection with Transformers](https://mlsanity.com/paper/2005.12872/) (2020)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Hierarchical Text-Conditional Image Generation with CLIP Latents](https://mlsanity.com/paper/2204.06125/) (2022)
- [Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding](https://mlsanity.com/paper/2205.11487/) (2022)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Bootstrap your own latent: A new approach to self-supervised Learning](https://mlsanity.com/paper/2006.07733/) (2020)
- [Flamingo: a Visual Language Model for Few-Shot Learning](https://mlsanity.com/paper/2204.14198/) (2022)
- [Parameter-Efficient Transfer Learning for NLP](https://mlsanity.com/paper/1902.00751/) (2019)
- [Prefix-Tuning: Optimizing Continuous Prompts for Generation](https://mlsanity.com/paper/2101.00190/) (2021)
- [Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision](https://mlsanity.com/paper/2102.05918/) (2021)
- [FiLM: Visual Reasoning with a General Conditioning Layer](https://mlsanity.com/paper/1709.07871/) (2017)
- [Visualizing and Understanding Convolutional Networks](https://mlsanity.com/paper/1311.2901/) (2013)
- [Visual Prompt Tuning](https://mlsanity.com/paper/2203.12119/) (2022)
- [Progressive Neural Networks](https://mlsanity.com/paper/1606.04671/) (2016)
- [Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learning](https://mlsanity.com/paper/1910.10897/) (2019)
- [Continual Learning with Deep Generative Replay](https://mlsanity.com/paper/1705.08690/) (2017)
- [Florence: A New Foundation Model for Computer Vision](https://mlsanity.com/paper/2111.11432/) (2021)
- [Robust fine-tuning of zero-shot models](https://mlsanity.com/paper/2109.01903/) (2021)
- [CURL: Contrastive Unsupervised Representations for Reinforcement Learning](https://mlsanity.com/paper/2004.04136/) (2020)
- [Lifelong Learning with Dynamically Expandable Networks](https://mlsanity.com/paper/1708.01547/) (2017)
- [R3M: A Universal Visual Representation for Robot Manipulation](https://mlsanity.com/paper/2203.12601/) (2022)
- [P-Tuning v2: Prompt Tuning Can Be Comparable to Fine-tuning Universally Across Scales and Tasks](https://mlsanity.com/paper/2110.07602/) (2021)
- [CLIPort: What and Where Pathways for Robotic Manipulation](https://mlsanity.com/paper/2109.12098/) (2021)
- [Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixels](https://mlsanity.com/paper/2004.13649/) (2020)
- [Learning multiple visual domains with residual adapters](https://mlsanity.com/paper/1705.08045/) (2017)

## Built on by (most cited first, 29 of 29)

- [RoboAgent: Generalization and Efficiency in Robot Manipulation via Semantic Augmentations and Action Chunking](https://mlsanity.com/paper/2309.01918/) (2023)
- [Gen2Act: Human Video Generation in Novel Scenarios enables Generalizable Robot Manipulation](https://mlsanity.com/paper/2409.16283/) (2024)
- [RoboCat: A Self-Improving Generalist Agent for Robotic Manipulation](https://mlsanity.com/paper/2306.11706/) (2023)
- [Affordance-based Robot Manipulation with Flow Matching](https://mlsanity.com/paper/2409.01083/) (2024)
- [TAIL: Task-specific Adapters for Imitation Learning with Large Pretrained Models](https://mlsanity.com/paper/2310.05905/) (2023)
- [Parameter-Efficient Fine-Tuning for Foundation Models](https://mlsanity.com/paper/2501.13787/) (2025)
- [Visual Tuning](https://mlsanity.com/paper/2305.06061/) (2023)
- [Polybot: Training One Policy Across Robots While Embracing Variability](https://mlsanity.com/paper/2307.03719/) (2023)
- [SpawnNet: Learning Generalizable Visuomotor Skills from Pre-trained Networks](https://mlsanity.com/paper/2307.03567/) (2023)
- [Parameter-Efficient Fine-Tuning for Pre-Trained Vision Models: A Survey and Benchmark](https://mlsanity.com/paper/2402.02242/) (2024)
- [Mitigating the Human-Robot Domain Discrepancy in Visual Pre-training for Robotic Manipulation](https://mlsanity.com/paper/2406.14235/) (2024)
- [Robust Finetuning of Vision-Language-Action Robot Policies via Parameter Merging](https://mlsanity.com/paper/2512.08333/) (2025)
- [Semantically Controllable Augmentations for Generalizable Robot Learning](https://mlsanity.com/paper/2409.00951/) (2024)
- [Diffusion Models For Multi-Modal Generative Modeling](https://mlsanity.com/paper/2407.17571/) (2024)
- [Dynamic Mixture of Progressive Parameter-Efficient Expert Library for Lifelong Robot Learning](https://mlsanity.com/paper/2506.05985/) (2025)
- [Efficient Continual Adaptation of Pretrained Robotic Policy with Online Meta-Learned Adapters](https://mlsanity.com/paper/2503.18684/) (2025)
- [Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Training](https://mlsanity.com/paper/2604.23121/) (2026)
- [Segment Any Events via Weighted Adaptation of Pivotal Tokens](https://mlsanity.com/paper/2312.16222/) (2023)
- [MResT: Multi-Resolution Sensing for Real-Time Control with Vision-Language Models](https://mlsanity.com/paper/2401.14502/) (2024)
- [GATS: Gather-Attend-Scatter](https://mlsanity.com/paper/2401.08525/) (2024)
- [Task-conditioned adaptation of visual features in multi-task policy learning](https://mlsanity.com/paper/2402.07739/) (2024)
- [CIVIL: Causal and Intuitive Visual Imitation Learning](https://mlsanity.com/paper/2504.17959/) (2025)
- [Factorizing Diffusion Policies for Observation Modality Prioritization](https://mlsanity.com/paper/2509.16830/) (2025)
- [π2vec: Policy Representations with Successor Features](https://mlsanity.com/paper/2306.09800/) (2023)
- [Understanding Representations Pretrained with Auxiliary Losses for Embodied Agent Planning](https://mlsanity.com/paper/2312.10069/) (2023)
- [Adapting Pretrained ViTs with Convolution Injector for Visuo-Motor Control](https://mlsanity.com/paper/2406.06072/) (2024)
- [Adaptive Policy Backbone via Shared Network](https://mlsanity.com/paper/2509.22310/) (2025)
- [Exploring Conditions for Diffusion models in Robotic Control](https://mlsanity.com/paper/2510.15510/) (2025)
- [RARA: Robust Policy Fine-Tuning through Retrieval-Anchored Representation Alignment for Robotic Manipulation](https://mlsanity.com/paper/2610.13578/) (2026)

## Abstract (arXiv)

Recent works have shown that large models pretrained on common visual learning tasks can provide useful representations for a wide range of specialized perception problems, as well as a variety of robotic manipulation tasks. While prior work on robotic manipulation has predominantly used frozen pretrained features, we demonstrate that in robotics this approach can fail to reach optimal performance, and that fine-tuning of the full model can lead to significantly better results. Unfortunately, fine-tuning disrupts the pretrained visual representation, and causes representational drift towards the fine-tuned task thus leading to a loss of the versatility of the original model. We introduce "lossless adaptation" to address this shortcoming of classical fine-tuning. We demonstrate that appropriate placement of our parameter efficient adapters can significantly reduce the performance gap between frozen pretrained representations and full end-to-end fine-tuning without changes to the original representation and thus preserving original capabilities of the pretrained model. We perform a comprehensive investigation across three major model architectures (ViTs, NFNets, and ResNets), supervised (ImageNet-1K classification) and self-supervised pretrained weights (CLIP, BYOL, Visual MAE) in 3 task domains and 35 individual tasks, and demonstrate that our claims are strongly validated in various settings.